^{Nothing in this article is a recommendation to transact in any financial product what-so-ever. The author of this article holds a ‘long’ position and stands to benefit if share prices rise of the company discussed here-in.}

Please read our terms of service.

One of the best investments you can make is in a business which is selling below historical multiples with no fundamental change in their balance sheet, and business operations all while it continues to return capital to shareholders and expand.

We believe CME Group Inc. (NASDAQ: ) presents a very attractive investment opportunity. It is a strong institutional business with record revenue, strong free cash flow, a healthy balance sheet, and a powerful exchange network.

CME is ‘the’ futures exchange.

Their equity index futures are the ones referenced on financial talk shows and checked by investors on Sunday nights. CME also offers markets on interest rates, foreign exchange, energy, agriculture, metals, and crypto.

Substack doesn’t allow free subscribers to utilize the AI Detection feature, we offer a screenshot and our AI Policy for your convenience.

The Spring Selloff

CME shares started to come under intense selling pressure in May after a fever dream of leadership changes combined with Kalshi announcing they would begin offering perpetual futures.

The market treated this announcement as a threat to CME’s moat. The stock now trades at at approximately 21.7x trailing earnings against their 12-month average of 25.1x well below it’s 3 and 5 year multiples, respectively.

Since the early-2020s, U.S. traders interested in perpetual futures had to go offshore, and in many cases use VPNs or other workarounds to bypass Know Your Customer (KYC) mechanics. If we view this holistically the underlying demand for perps was already out there; Kalshi’s launch simply allowed the demand to remain onshore.

The launch did not create a whole new TAM, but simply allowed the trading volume to remain inside the U.S. This matters greatly, because if perps were truly a major threat to CME’s business model, the damage would have likely shown up earlier in the stock’s share price.

As such,

We believe the threat to CME’s core business to be vastly overestimated and misunderstood by the market; and our argument is backed by the fact that CME is continuing to produce near-record volume even after the launch of perps.

• Kalshi proved there is retail demand for U.S.-regulated perps, this demand has existed for years and does not mean CME’s main business is at the risk reflected in the share price.

• CME’s own Q2 remarks stated 94% of first-half 2026 volume came from institutional customers. As stated above Perps are primarily a retail oriented product and are simply not appealing to institutional traders.

  • Perps are highly leveraged, volatile, and have delicate funding mechanisms with significant liquidation risk which are not appealing to the risk appetite of institutional traders.

    

The question is whether the fundamentals moved with the multiple. The Q2 print says they did not.

CME-Q2-2026 Breakdown

• Revenue: $1.706.2B, +1% YoY

• Adjusted EPS: $2.99, +1% YoY

• Market data revenue: $238M, +20% YoY

• ADV: 29.8M contracts, within 1% of record Q2 last year.

• Open interest: +8% YoY.

Read more

Jersey Mike’s consists of approximately 3,300 franchised stores that make sandwiches, and Jersey Mike’s collects 6.5 cents of every dollar they sell, plus advertising fees. It owns only 26 stores; and spends next to nothing on capital. The business is primarily a royalty collection machine.

First, Jersey Mike’s is a royalty collector, so in a normal year cash from operations should track Adjusted EBITDA closely as there’s no inventory, no store capex, and no working capital drag to speak of. Combined Adjusted EBITDA for fiscal 2025 was $339 million.

However, instead it reported negative $210 million, because of where the founder’s $411 million in transaction bonuses landed on the cash flow statement.

Under GAAP, bonus payments to employees are compensation, and compensation is an operating activity. It doesn’t matter that the bonuses were triggered by the Blackstone sale and will never recur; the cash went out through the operating section, not financing or investing.

Blackstone bought into Jersey Mike’s on January 16, 2025. On July 2, 2026, the company filed its S-1 to list on the NYSE under the ticker JMKE.

The headline numbers:

• Over $4 billion in systemwide sales.

• Revenue of $724 million in fiscal 2025, up 11% from $653 million.

• Adjusted EBITDA of $339 million.

We believe three other things in the filing deserve more attention:

1. Where did twelve points of margin came from in two years?

2. Why did the company issued $760 million of debt five months before IPO?

3. What “50% cumulative same-store sales growth” means once you read the footnote.

‘Explosive’ Margin Growth

Adjusted EBITDA margin was 35% in 2023. In fiscal 2025 sky-rocketed to 47%.

Adjusted EBITDA, for readers newer to filings, starts with earnings and adds back whatever management defines as non-recurring or non-operational. Every company writes its own definition. Jersey Mike’s has nine categories of add-backs, and the margin story lives inside them.

The first source of margin is the founder. One of the nine add-backs is “founder-related discretionary expenses,” described as founder-directed bonuses and charitable donations not expected to recur under Blackstone. These declined $181 million year over year in 2025.

Separately, the founder paid $411 million in transaction bonuses at closing, and those bonuses ran through operating cash flow. Fiscal 2025 operating cash flow was negative $210 million as a result. Over the same period, the S‑1 states that the business converted approximately 97% of Adjusted EBITDA less capital expenditures into ‘free cash,’ as a non‑GAAP management metric.

Subscribe now

There is also an airplane. In 2024 the company bought a $41 million aircraft on the founder’s behalf and transferred it to him in connection with the sale.

The second source is advertising. Franchisees pay a percentage of sales into a national ad fund, and that fund is meant to cover the marketing budget. Under founder ownership it didn’t; the company topped it up with its own money, $48 million in 2023 and $38 million in 2024. Blackstone ended the subsidy. The company now spends strictly what the fund collects.

The third source is the Area Director program. Jersey Mike’s historically paid independent Area Directors roughly 2% of gross sales in their territories to recruit franchisees and support stores in the field. At current systemwide sales, that program was worth about $80 million a year at full scale. Blackstone has been aggressively buying out the contracts. As of May 2026, one Area Director remains, covering about 1% of the system.

Make note of the accounting treatment, because it repeats. The Area Director buyout costs are excluded from Adjusted EBITDA as non-recurring. The savings from the terminated contracts are not excluded; they are the margin. The same asymmetry runs through the add-backs generally: costs of the transition are adjustments, benefits of the transition are profit.

The headline for IPO investors is simple: most of the dramatic Adjusted EBITDA margin expansion from 2023 to 2025 comes from discrete expense cuts—founder discretionary spending, advertising subsidies, and Area Director fees—rather than deep structural changes to the business, and those levers now appear largely exhausted, which we believe presents a more challenging runway for margins and growth into 2026 and beyond.

Jersey Mike’s reported negative $210 million in operating cash flow last year. Same store sales are staling, while the same filing claims its business converts 97% of Adjusted EBITDA into cash. More below.

Read more

This article is covered by our Terms of Service, which you can read here.

We are continuing our analysis of CrowdStrike ( ) after the stock sold off despite an earnings beat.

Going into the Q1 FY27 print on June 3, CrowdStrike traded at roughly 154 times forward earnings and around 37 to 40 times trailing revenue. This left very little room for error on the stock, and the market expected a much stronger print to justify the valuation.

After the print, sold off aggressively afterhours (retreating 11% at one point.)

But there is a cost buried inside of CRWD’s income statement that incubates and slowly erodes earnings and silently dilutes current shareholders.

Stock-based Employee Compensation

CrowdStrike has a very aggressive SBC program for employees, and it’s a bill that shareholders pay in the fog-of-war that is GAAP Vs. Non-GAAP reporting. When CrowdStrike pays a software engineer, or sales director in stock instead of cash, Generally Accepted Accounting Principles (GAAP) mandates CRWD to record that as an expense, in the same way it would salary disbursements. Non-GAAP on the other hand does not consider this cash leaving out the door, so it’s stripped out and presents it as a much cleaner profit number.

CrowdStrike SBC ran $297.7 million for the quarter, and $317 million once we throw in employer payroll taxes. That equates to approximately 23 percent of revenue being…

Redirected to employee equity.

One of every five dollars CrowdStrike earned went right back to employee equity before GAAP profit was tallied. Non-GAAP considers this aggressive compensation to cost the company nothing.

Share

A Ticking Time Bomb

But for now, the can is kicked down the road. Net new ARR is up 32 percent. The stock is up 97 percent off the April low (a cyclic short we called in our Short Thesis.)

When price appreciation runs faster than dilution, the non-GAAP story holds and nobody asks hard questions about the cost structure underneath it.

The concern is what happens at the other end of the growth curve.

The ARR base is now $5.51 billion. Compounding at 32 percent gets harder every year, not because CrowdStrike is losing, but because the math of large numbers wins eventually. When growth normalizes, the multiple compresses. A stock that deserved 154 times earnings during hypergrowth does not deserve that after it. Compression is how markets correct for optimism.

And that is when the SBC structure becomes a trap.

If you’re a CrowdStrike shareholder (non-employee) this should be your top concern.

Read more

In this case, Entravision — — is misclassified by the market as a legacy Spanish broadcaster. While, yes, technically they are in-fact a legacy broadcaster which is merely, in our opinion, their margin of safety behind their booming mobile ad-platform; which generated 154 million in Q1-2026, and equated to an impressive +204% year-over-year. The company’s legacy media assets provide stability, but the investment case is increasingly driven by its fast-growing digital advertising platform. Their digital advertising platform is built on a proprietary AI system which we believe is still grossly undervalued by the market.

But first, you must understand why EVC is an AI platform, that happens to be in legacy broadcasting—

A machine learning titan before the AI boom.

EVC owns Smadex and Adwake which are the core to EVC’s Advertising Technology and Services (ATS) segment. Combined, they generated 270 million in full year 2025 revenue, and then proceeded to nearly match that number in Q12026 alone.

It is very important for investors to understand what Smadex and Adwake does as they both represent the diamond in the rough the broader market has overlooked in a fever dream over AI-chatbots.

Smadex is a mobile device centric demand side platform (advertising.) It’s software that functions as the brain that sits on the advertiser side of the digital ad auction market. Smadex in real time computes calculations for millions of ad slots across mobile apps, and then predicts which users are most likely to do what (install, purchase, register etc.) and Smadex places bids automatically. At the core is a robust, proprietary machine learning engineered trained to predict the future of a user (which sounds a-lot like we do, but with earnings calls.)

Subscribe now

Adwake on the other side of the coin is the managed services and growth layer built on top of Smadex. Where Smadex is the plane, Adwake is the pilot, navigator, and flight-crew. Adwake runs advertising campaigns for clients across gaming, fintech, e-commerce etc.

We can not stress enough, that the combination of Smadex and Adwake forms a fully organic Machine Learning-AI (ML-AI) company hidden inside a broadcaster. There is still tremendous upside for this stock that the market still has not priced in.

Truly organic AI companies are RARE.

Most self described “AI-Companies’ are in-fact, not AI companies. They’re AI wrappers, they consume tokens, and pass queries to the likes of Anthropic and OpenAI. Which makes them no different than a toll booth built on rented land. The toll booth operator owns NO property, and anyone can do it.

LLM inference costs account for 80-90% of total on-going cost of using AI, and is one of the largest margin strains for AI-Companies. The problem starts to snowball with their rate of revenue. Which in turn compresses margins and reduces operational leverage. In short, for the toll-both operator, the rent’s going up.

BUT—EVC doesn’t have this problem.

Because Smadex is the polar opposite of this: according to their own published engineering documentation, Smadex processes one million bid requests per second, each requiring a complete AI-driven bidding decision within a 100-millisecond window. At that latency requirement, an external API call is physically impossible — the round-trip network latency alone would consume a meaningful fraction of the entire budget. Every inference is executed internally, on proprietary models, on infrastructure that Smadex controls

Smadex’s in-house ML models are performing several simultaneous computations:

• Conversion probability prediction: What is the likelihood this user, in this context, will complete the target action (install, purchase, registration)?

• Bid price optimization: Given predicted value, what is the optimal bid that maximizes expected return rather than just win rate?

• Budget pacing: How should this bid interact with campaign pacing, frequency caps, and daily budget constraints?

• Creative selection: Which ad format and creative is most likely to drive the desired outcome for this specific user?

All of this is running on their own, proprietary deep learning models that Smadex’s in-house ML team trains.

The key insight for all of this is: Smadex (EVC) does not outsource anything. There is no OpenAI API Call, there is no per-hit cost to a third party inference vendor. The cost of running this is embedded into Smadex’s cloud infrastructure via AWS. Even if you own the land, and the toll booth, you still have to pay the power company.

If we look at a similar company running something equivalent on a non-native stack even at generously discounted enterprise rates; we watch margin health rapidly deteriorate. Smadex and EVC do not have this problem. Because Smadex was built from the ground up. EVC owns the land, and EVC owns the toll booth (Smadex, Adwake)

Agentic Margins

We reviewed a Smadex Machine Learning Engineer job posting and can deduce management is fully aligned with using Agentic-AI as a high fulcrum growth lever. Most companies, eat into their margin health with run away AI spending in an attempt to keep up with the times. EVC will not have this problem, because machine learning and advanced neural networking is the core of their business.

Special Forces soldiers are often referred to as ‘force-multipliers.’ In a holistic sense, we view agentic engineering teams in a similar lens as they produce more output per headcount than traditional engineering teams. Independent benchmarks from enterprise deployments in 2026 show that agentic coding tools reduce software development costs by 20–45% and deliver a code review agent completing a routine pull request for $0.72 versus $48 of senior engineer time — a 66x cost reduction on standardized work.

Applying this directly to Smadex: their engineering headcount is being hired specifically to build and maintain automated ML systems. Each hire is effectively operating at a multiplied output rate compared to a traditional engineer. Management noted in the Q1 2026 call that they "invested in our engineering team to continue to improve our technology and build more powerful AI capabilities into our platform" and simultaneously increased operating leverage. The two facts are connected: the agentic architecture of their engineering team means that adding engineers does not add engineering costs proportionally — because each engineer is running ML pipelines that automate large portions of their own workflow.

We believe ATS revenue will comfortably outpace ATS engineering costs over the next three years, which in turn will improve margin health—because the team is building a system that compounds their own productivity and they are doing it with an already professional coded, purpose built, in-house system. Most companies are building their AI-car as they drive it. But EVC built the car 5 years ago, and now all management has to do is a drop-in upgrade.

What Is EVC Actually Worth?

We are not applying aggressive assumptions to get there. We are not extrapolating the 204% growth rate forward indefinitely. We are simply asking: what would a standalone, profitable, ML-native programmatic DSP doing $650+ million in annual revenue trade for if the market knew what it was looking at?

The answer is: not $7.

Based on ATS segment economics, with 154.6 million in Q1 revenue and 23.3 million 34.3 million in operating profit as our base, we applied a conservative ad-tech EBIT multiple discounted relative to a comparable public company AppLovin ( )

We come to a base case conservative price target of $12.00

We say conservative because in this model we are modeling the legacy broadcast business (FCC Spectrum, Political Advertising) at ZERO.

For now, this is enough for you to understand why EVC may not look like an AI-forward, and efficient DSP on the surface, but it shows the deeper you dig into the company. And the market still has only scratched the surface for this stock.

EVC shareholders had a preview of what happens when the market starts pricing EVC like the ML powerhouse it actually is instead of the sleepy broadcaster it appears to be on the surface.

In our next EVC post we’ll cover our Bear/Base/Bull scenarios and map ATS operating profit more specifically.

Credits

A special thanks to for setting the foundational coverage for this stock on Substack. I’ve been long the stock based on their analysis going back to late 2024. You can read his coverage here:

The Value Road

The Hidden Asset Just Got Louder

In November I wrote that EVC was being priced like a dying broadcast relic when it was actually one of the fastest-growing ad-tech platforms in the world, attached to a Spanish-language broadcast footprint heading into a 2026 political cycle, sitting on FCC spectrum the balance sheet won’t mark to fair value. The stock ripped 48% the next day on Q3 earn…

Read more

3 months ago · 16 likes · 1 comment · The Value Road

For new readers: Kalshi’s keyword mention contracts for earnings calls are event contracts regulated by the CFTC. They resolve based on whether management, like — Tim Cook and CFO Kevan Parekh or call operator(s) says a specific word during the call itself.

Call Structure

Apple’s earnings calls are the most scripted in big tech. Our review of 83 transcripts spanning 24 years shows a pattern that borders on mechanical.

The format rarely varies. Cook delivers prepared remarks touching each product segment in order — iPhone, Mac, iPad, Wearables, Services — then pivots to values, geopolitics, and macro positioning. Parekh follows with a financial walk: segment revenue, gross margin, guidance. Then Q&A, where eight to ten analysts from bulge-bracket firms ask pre-telegraphed questions they have been publishing in preview notes all week.

Cook is one of the most disciplined call operators and his linguistic fingerprint is just as consistent. He almost never introduces vocabulary that wasn’t in his prepared script. He does not name unannounced products. He deflects hardware roadmap questions with one rehearsed phrase: “We don’t comment on products we haven’t announced.”

The Last Dance

On April 19 — eleven days before this earnings call — Apple announced that Tim Cook will step down as CEO on September 1, handing the role to hardware chief John Ternus.

This is one of Cook’s final two earnings calls as the person who ran the most valuable company in history for fifteen years. That changes the tenor of everything.

Outgoing CEOs in their final quarters do not experiment with new language. They do not float new narratives or speculate about long-dated products. They do the opposite. They anchor to legacy, stabilize investor confidence, and deliver the story they want told on their way out. For Cook, that story has three pillars: the Services empire, the Apple Silicon transition, and the supply chain discipline that produced $600 billion in U.S. manufacturing commitments.

The Free Look: FOLDABLE Is The Cleanest NO On The Board

The foldable iPhone — rumored as “iPhone Fold” or “iPhone Ultra,” expected September 2026 — is the most-covered Apple story in tech media right now. Bloomberg, The Information, Ming-Chi Kuo, and Mark Gurman have all written extensively about it inside the last 30 days. The press coverage is everywhere. The management language is nowhere.

While we believe there is a chance an analyst may say something to the effect of

“Could you tell us about the ongoing rumors about a foldable iPhone?”

We expect Cook, or any other company rep to say the usual…

“We don’t speculate on future products.”

Cook has never named an unannounced product on an earnings call. Not in 24 years. Not once. Analysts who ask get the same deflection line every time. An outgoing CEO protecting his legacy has even less incentive to make speculative hardware promises he won’t be around to deliver on.

Our model has this contract scoring in the single digits. Wherever the market sits on this one, the directional read is clear: foldable does not appear in management’s spoken language on Thursday.

The same logic applies to GLASSES, with even stronger force. Apple’s AR/smart glasses are not expected until late 2026 at the earliest, with some estimates pushing into early 2027. There is no shipped product, no confirmed spec, no name. Press coverage is loud. Apple’s investor communications are silent. In over eight decades of cumulative earnings call transcript history, Apple has never named a glasses-category product before launch.

Both of these are the same trade structurally — the news-cycle volume is doing the work in market pricing, and the management discipline is doing the work in our analysis.

POSITION DISCLOSURE:

The author of this article holds the following positions:

• kxearningsmentionsaapl — “FOLDABLE” — NO

• kxearningsmentionsaapl — “GLASSES” — NO

Read more

Kalshi traders are pricing Zyn US shipment volume for Q1 2026 at an implied median at or about 183 million cans. PMI reports the number on April 22. Based on our own research and modeling we believe Zyn will ship a number north of 183 million cans.

PMI’s own 2026 guidance calls for +15 to 20% full-year growth. Nielsen offtake ran +23% YoY in Q4 2025. Haypp’s US pouch category grew 35% in January. PMI’s CFO confirmed on the Q4 call that 25M cans of surplus channel inventory still need to unwind. Q1 is historically a build quarter, not a draw.

Subscribe now

Every primary source on the buy side, sell side, and scanner side points up. The market is pricing down. Below is our thesis:

Read more

We’re coming off a big win on Nike’s earning call. The Truist strikes are new for our machine learning model, and our first analysis of a bank.

Truist Bank is the 7th largest bank in the nation with over 1900 branches nationwide. Their call is scheduled for pre-market on Friday April 17th it will be broadcasted here.

We believe their prepared remarks will stress heavily: revenue growth, NIM drifting upward, and managements path to 15% ROTCE by 2027. Driven by heavy investments into mid-market lending, investment banking, and AI-integrations.

Kalshi Traders currently have the odds of management talking about the following topics set at:

We ran every word listed above through our ML probability model, context classifier, and social research overlay.

Below is our research based on each strike listed above.

TAILWIND at 51¢ is the most underpriced contract on this board

Two recent Truist calls had management use the word unprompted.

In Q4 2025, Ebrahim Poonawala (BofA) asked about rate sensitivity and NIM in a lower-rate environment. Rogers responded with the phrase “natural tailwind” when describing QE and lower rates for the industry.

In Q3 2025, Ken Usdin (Autonomous) pressed on fixed-rate asset repricing timelines. Maguire described the repricing effect as a “tailwind” for NIM as higher-yielding assets replaced maturing low-yield positions. Again, his framing, not parroting.

Two different executives. Two different analysts. Same word. Both tied to rates and NIM.

Why Q1 should repeat the pattern:

The Fed is on hold. Repricing dynamics are front and center for banks this quarter. Truist’s own commentary and sector previews have NIM trajectory and balance-sheet positioning as top discussion topics.

Poonawala and Usdin are both on the Truist coverage list and almost certainly get questions again. Poonawala reliably asks about rate sensitivity and structural NIM. Usdin is the dedicated repricing/NIM mechanic. When either asks their usual question, the most natural language Rogers or Maguire have reached for historically is exactly “tailwind” — either “natural tailwind” or “repricing tailwind.”

• Management at multiple banks (JPM, regional peers) has used “tailwind” in NIM/NII and balance sheet discussions.

• Sector commentary going into this earnings week repeatedly frames loan growth and NIM as tailwinds for the group.

• Rogers used “natural tailwind” for the industry in Q4 2025 when Poonawala asked about rate sensitivity.

• Maguire used “tailwind” in Q3 2025 to describe fixed-rate asset repricing when Usdin pressed him.

Our conclusion: “tailwind” is currently common bank jargon for exactly the topics Truist will be asked about on this call (NIM, repricing, NII). Recent peer calls show both analysts and management are comfortable using it, which materially boosts the probability Truist management will say it again when similar questions arise this week.

That’s the free look. The full scorecard — including our highest-conviction position with, all eleven keyword analyses, and the complete analyst-by-keyword exposure map — is below for subscribers.

Read more

On March 16, the CFTC published an Advance Notice of Proposed Rulemaking on prediction markets (RIN 3038-AF65).

The comment period is open. Any member of the public can submit a comment on the federal record.

Sadly, most comments are boiler plate blogs of text about how Prediction Markets are bad, evil, and scary. They’re devoid of any rational argument, and are rooted in feelings not fact; and there are dozens of copy and pasted comments like the one below. They use scary terms like terrorism, and war; in an attempt to sell a narrative that Kalshi traders are running around betting on coups and bloodshed.

Except, they aren’t.

They can’t.

Because these type of contracts are already prohibited under the law.

In an attempt to provide some more color beyond individuals spamming the public record with copy pasta. Today, we submitted a formal public comment raising five issues we believe no other commenter has raised in the docket. The full comment is linked at the end of the article. This is the short version

Subscribe now

What we propose

1. A prediction market Self Regulatory Organization (SRO.)

Every other regulated asset class in the U.S. has at least one self-regulatory organization between the exchange and the federal regulator. Equities have FINRA. Municipal bonds have the MSRB. Prediction markets have nothing. The NFA has no jurisdiction over exchanges — only over intermediaries like FCMs

The CFTC is supervising this entire asset class alone, directly, with no buffer. We proposed a structurally independent PM-SRO, modeled on the precedents Congress set when it created the MSRB in 1975 and consolidated FINRA in 2007. A prediction market only SRO would allow disputes to be resolved outside of Kalshi’s current format of dispute resolution. We believe this would stand to benefit traders greatly.

And a point we think is one of the most important:

The current professional exam which governs event contracts (Series 3) contains zero questions about event contracts.

Event contracts are very nuanced, settlement mechanics, listing, disputes and other variables can be confusing and highly technical. A registered commodity professional can pass the Series 3 and know nothing testable about the fastest-growing product class on CFTC-regulated exchanges. So we proposed a dedicated “Series PM” competency exam.

This is like if the bar exam covered every area of law except criminal defense, but passing it still licensed you to represent murder defendants. The courthouses are full. The caseload is exploding. And every public defender walking in the door passed a test that never once asked them what “beyond a reasonable doubt” means.

2. Prohibit branding between registered and unregistered platforms.

At least one operator runs a CFTC-registered domestic platform and an unregulated offshore platform under the same core brand name. In early 2026, a contract on the death or removal of the Supreme Leader of Iran resolved differently across these two platforms. Same brand. Different outcomes. The domestic platform cancelled the contract. The offshore platform paid out. We proposed mandatory branding separation, point-of-sale disclosure, and a prohibition on cross-platform promotion of unregistered affiliates.

Dual branding of regulated + unregulated platforms confuses the general public, especially with markets relating to events which are banned under the CEA.

3. Mandatory algorithmic trading disclosure.

The whole policy argument for prediction markets is that prices reflect distributed human judgment. That argument falls apart if an undisclosed share of volume comes from bots running arbitrage and market-making strategies. We proposed a framework modeled on the CFTC’s own previously proposed Regulation AT: registration of automated trading systems, monthly publication of aggregate automated-volume statistics, and strategy-category disclosure. No proprietary parameters exposed. Just enough transparency to know whether the “wisdom of crowds” is actually a crowd.

4. Standardized public API and data reporting.

Prediction market prices show up in news broadcasts, policy briefings, and intelligence assessments. No two registered platforms publish data in the same format. You can’t do cross-platform surveillance if every exchange speaks a different data language. We proposed a standardized public API, five-year historical archives, and daily automated surveillance reports — analogous to the Consolidated Audit Trail for equities. If this data is going to be treated as public information infrastructure, it should meet a public information standard.

Why we filed

First Strike Research is a financial research operation. We trade prediction markets. We study them. We’ve built tools to analyze them. We have a direct commercial interest in these markets functioning with integrity, and a direct analytical interest in their data being accessible and standardized.

We also think the CFTC is at an inflection point. The Commission asked the public for input. Most of the comments in this docket, from what we’ve seen, argue about whether prediction markets should exist or how wide the product scope should be. Those are fine questions. But the plumbing questions — who watches the exchanges, what data standards apply, how do you tell bots from humans in the order book — are where the actual regulatory failures are happening right now. That’s what our comment addresses.

Read the full comment

Our complete comment, in PDF Format, with legal citations and proposed remedies for each section, is linked here.

First Strike Research

outreach@firststrike.trading

First Strike Research is not registered as a Commodity Trading Advisor (CTA) with the NFA or CFTC. Past performance figures cited in this article, including win rates, directional accuracy percentages, and P&L results, are historical and are not necessarily indicative of future results. Hypothetical or simulated performance results have inherent limitations and do not represent actual trading by any specific subscriber. Individual results will vary. Subscriber testimonials reflect individual experiences and are not a guarantee of future performance. Prediction market trading involves substantial risk of loss, including the total loss of invested capital.

Our newest model just went 8-for-10 on traded positions across Nike’s Q3 FY2026 earnings call Kalshi keyword contracts, that’s an 80% win rate on traded positions, and 88% directional accuracy across all 16 contracts analyzed

And the part we’re most proud of?

Our two highest-conviction trades both hit. If you would’ve traded this alone you would’ve netted a 200% ROI.

These were positions where our ML probability model and our social research overlay converged and were in full agreement with the highest possible conviction score.

Don't miss out, subscribe now.

But don’t just take my word for it, our paid subscribers took down some handsome profit—

The Three-Layer System Is Getting Sharper Every Cycle

We’ve now run our new architecture through two complete earnings cycles — CrowdStrike and Nike — and the trajectory tells a clear story.

CrowdStrike: 11/13 directional (85%)

Nike: 14/16 directional (88%), +208¢ net P&L

That’s not a fluke. That’s a system learning and improving in real time.

Layer 1 — The ML Model generates raw probability estimates based on historical transcript data and linguistic patterns. It sets the baseline. On Nike, the raw model alone tracked around 10 out of 16 — decent, but not where we want to be. Its strength is consistency: it processes every keyword through the same statistical lens and catches patterns that gut instinct misses.

Layer 2 — The Context Classifier overlays narrative lifecycle analysis. Is a term rising in management’s vocabulary or fading out? This layer caught that Consumer Demand and Amazon were losing narrative momentum even though the model’s base rates still liked them. It correctly flagged Express Lane and Nike Factory as terms drifting out of active rotation. Layer 2 brought us from roughly 10 to 12 out of 16.

Layer 3 — The Social Research Overlay is where the edge lives. Deep research into transcript history, brand strategy signals, analyst commentary, media coverage, and social sentiment catches the event-driven shifts that data alone can’t anticipate. One of our biggest social research win was with Starbuck’s Condiment Bar.

Three critical interventions on Nike show why this layer changes everything:

The Tariff flip saved 97¢. Our model had this as a BUY NO at 63¢. The market had it at 97¢. Our research identified that in the current macro environment, Nike management was almost certainly going to address tariffs head-on. We flipped the position. That single research-driven call was worth nearly a dollar per contract.

The Retailer flip earned 67¢. Research showed wholesale distribution dynamics had become a hot topic in the analyst community — something the model’s historical data underweighted. Our social and media overlay caught the shift.

The Dividend risk flag limited a loss. We identified the April 1 pay date and Nike’s stock sitting near a five-year low as catalysts that could force the word into Q&A. We sized down accordingly. The loss came in at -43¢ instead of what could have been significantly worse.

A Tale of Two Misses

We’re celebrating the wins, but it’s important to remember that no model, formula, or system will always be 100% accurate.

Holiday (-81¢) was our biggest miss - Nike’s Q2 call was the holiday quarter — management had already spent extensive airtime on seasonal performance. By Q3, they’d moved on. The model liked Holiday based on historical Q3 base rates, but the word was temporally anchored to the prior quarter. In the future, we’ll more likely than not advise NO ACTION or ‘NO’ contracts.

Dividend (-43¢) hurts less because our research actually identified the risk. We sized down instead of flipping to a full PASS or even YES. Going with the gut here, YES contracts would’ve been more appropriate given existential share price retreat.

The Scoreboard Speaks for Itself

Two earnings cycles with the new model are now in the books, and here’s what the data says about this system:

Improving accuracy — 85% to 88%. Positive and growing P&L. Highest-conviction picks landing at a perfect clip. And every miss traceable to a fixable process gap rather than a fundamental flaw in the methodology.

The ML model provides the statistical backbone. The context classifier reads the narrative landscape. The social research overlay captures the real-world signals that move markets. When all three layers agree, we haven’t been wrong yet.

First Strike Research publishes analytical commentary on prediction market contracts. This content is for informational purposes only and does not constitute a trading recommendation, financial advice, or an offer to buy or sell any contract. First Strike Research is not registered as a Commodity Trading Advisor (CTA) with the National Futures Association (NFA) or the CFTC. Prediction market contracts on platforms such as Kalshi may constitute commodity interests regulated under the Commodity Exchange Act. Trading event contracts involves substantial risk of loss, including the total loss of invested capital. Past model performance (e.g., 10/13 on CrowdStrike) is not indicative of future results. The author may hold positions in the contracts discussed above at the time of publication. Nothing in this publication should be construed as a recommendation to enter into any specific trade.

Terms of Service

Prediction markets for earnings call keywords are an interesting corner of the rapidly expanding prediction market landscape. Most traders anchor to historical base rates, and nothing else. If you look at the comments section of any Kalshi contract most traders are only going back 1 quarter.

We’re attempting to use as much technology as possible and find and deliver to you advanced insight, and we’ve invested significantly into training and tailoring our model including using Claude’s-API to read around tradeable words to understand the given context and score the odds accordingly.

Our new model went 10/13 on the CrowdStrike call.

This is our first post sharing our new fair value model on Nike’s Q3 FY2026 call which is Tuesday evening. We’ve run every strike in our model against current intelligence: recent earnings transcripts, analyst previews, press releases, product launches, and macro developments from the last 90 days.

We’ll start with the 2nd highest conviction trade free for all subscribers. Our highest conviction play, plus the full scorecard on every remaining strike is available ONLY to our paid subscribers.

2nd Highest Conviction Trade: EXPRESS LANE

Our Model: 47¢ | Market: 66¢ | Our Position: NO

This is a 19-cent edge on a contract we have high conviction the market is overpricing. Here’s the full reasoning.

What Is “Express Lane”?

Express Lane was Nike’s mid-2010s manufacturing innovation program — a supply chain acceleration framework designed to cut product time-to-market from 18 months down to a matter of weeks. Under CEO Mark Parker, it was a centerpiece of Nike’s innovation narrative and a recurring fixture in earnings call language.

It was genuinely exciting at the time. It signaled agility, responsiveness to consumer trends, and operational sophistication. Management mentioned it repeatedly. Analysts asked about it. It had a real moment.

Express Lane’s time in the spot light has come and gone.

The Graveyard

Corporate communication follows predictable lifecycle patterns. Keywords get introduced, peak, plateau, and then — when the strategy they represent is superseded — they get retired. Not with an announcement. Just quietly, by omission. Management stops reaching for the phrase. Analysts stop referencing it. It fades.

Express Lane is deep in this fade cycle. The phrase does not appear in Nike’s Q2 FY2026 earnings transcript. It does not appear in analyst preview notes for Q3. It is not referenced in any Nike press release or investor communication we can find in the last 12 months. The current turnaround vocabulary — “Win Now,” “Sport Offense,” wholesale partner relationships, brand heat — has completely displaced it.

More importantly: the person who built Express Lane isn’t running Nike anymore. Elliott Hill took over in October 2024. His operational framework is entirely different. He’s not going to reach back to a Mark Parker-era initiative to describe what his supply chain team is doing.

What The Market Is Missing

The 66¢ pricing implies a roughly 2-in-3 chance that management uses this specific phrase on Monday’s call. We think that’s dramatically wrong.

When we search for any evidence of Express Lane appearing in Nike’s recent communications — investor presentations, newsroom releases, leadership interviews — we find nothing current. The most recent substantive coverage of the program dates to mid-2010s trade press.

For a word to appear at 66¢ probability, it should be woven into the current narrative. It should be in analyst notes as something to watch. It should be part of the company’s active vocabulary. Express Lane is none of those things.

Our fair value estimate: For YES is at ~20-25¢. The market has YES priced at 66¢ represents roughly a 40-45¢ overpricing. Our NO position at 47¢ captures a real but conservative slice of that edge — leaving room for uncertainty and the small chance an analyst asks a legacy question that pulls the phrase out.

Sources for Express Lane analysis: Nike Q2 FY2026 earnings transcript (Dec 18, 2025); Nike newsroom press releases (Q1-Q3 FY2026 window); analyst Q3 preview notes from BofA, UBS, Seeking Alpha, Benzinga; Nike investor relations presentations; FashionNetwork archival coverage of the Express Lane program (2016-2017).

PAID SUBSCRIBER ONLY CONTENT BELOW

What follows: our highest conviction trade of the entire board (with a 13-18¢ edge we think is being left on the table), plus our full intelligence overlay on Tariff, Supply Chain, Holiday, Sabrina, Dividend, Nike Factory, Retailer, Consumer Demand, China, Sustainable, Direct to Consumer, Amazon, and Digital Sales. Every keyword, every data point, every recommendation.

Read more

Coming off a whiff on our recent high conviction play, we went back and rekeyed our machine learning model; we added AI-Enhanced abilities that provide context to the actual words spoken and the new results are simply impressive.

CrowdStrike reported earnings on March 3rd. Kalshi listed 13 binary contracts on whether specific words would be mentioned during the call. Accept our apologies, as with the new model weights we were not entirely comfortable releasing our predictions to readers before we could validate the new system.

The model nailed 11 out of 13 directional calls. Every high-conviction bet was correct. Every fade was right. The only misses were two event-driven surprises that no historical model could have predicted and we caught one of those with a manual research overlay before the market opened.

Here’s exactly how we’ve improved our model, which we will be utilizing to provide fair values models on contracts to our readers going forward.

The Problem With How People Trade Mention Contracts

Most traders on Kalshi approach earnings mention contracts the same way: they pull up the last transcript, ctrl-F the keyword, and make a gut call. Maybe they check two or three transcripts if they’re diligent. The market price reflects this — a loose consensus of people eyeballing recent history.

That approach has two blind spots. First, it doesn’t weight recency properly. A keyword mentioned 8 of the last 8 quarters feels like a lock, but if 6 of those mentions were three years ago and only 2 were recent, the probability is very different than if all 8 were consecutive. Second, and more importantly, it treats all mentions equally. “We are investing heavily in our consolidation platform” and “as we’ve moved past the outage incident” are both keyword hits. But the first one is a CEO describing core strategy. The second is a CFO giving a one-sentence brush-off to an analyst’s question. One predicts future mentions. The other predicts silence.

We’ve made a tenfold technical improvement to our model, and the new and improved system moves to addresses both problems.

The Results

Our model correctly forecasted the following:

Correct YES calls (6/6):

• Falcon Flex at 95¢

• Consolidation at 93¢

• AWS at 75¢

• Hyperscaler at 73¢

• Nvidia at 68¢

• Acquisition at 60¢.

Correct NO calls (5/5):

• Outage at 45¢

• Generative AI at 35¢

• China at 28¢

• Hack at 12¢

• Shadow at 40¢

The two INCORRECT calls:

Both Microsoft and Signal model forecasts were incorrect, however, upon human review we more likely than not would’ve published these strikes as “No-Action” (e.g., do not trade it.)

The full technical breakdown of our new two-layer model, including how the LLM context classifier works, the seven adjustment signals, and the complete strike-by-strike classifications. The technical details are available below for paid subscribers.

Going forward, this is the system we’re using to generate fair value estimates on every Kalshi earnings mention contract. Paid subscribers receive our probability tables and trade signals before the market opens. Free subscribers will receive analysis on one or two keywords we decide to publish publicly.

We went 11/13 on CrowdStrike; and the next cycle is already loading.

Subscribe now

Paid subscribers, once more thank you for your patronage. Your subscription directly supports the financial obligations required for our machine learning system (API calls, LLM Tokens, etc.) In step with our commitment to continue providing a strong value proposition, please enjoy reading about how we’ve improved the model to help support your trades. And, in the interest of your time the following is very technical, if you’re interested in how the model you pay for actually works, read on!

How the Model Works: Two Layers

Layer 1: The Frequency Model

We trained a random forest classifier on 3,144 historical earnings call outcomes across seven public companies: CrowdStrike, Google, Meta, AMD, Costco, Tesla, and Goldman Sachs. We are actively training the model on more relevant earnings calls, and companies. The current training data spans roughly 12 years of transcripts.

For each company and keyword pair, we extracted 43 features from the historical record. The model learned which patterns predict a YES and which predict a NO. It achieved 80.4% accuracy in cross-validation with an AUC of 0.881 — meaning it correctly ranks higher-probability keywords above lower-probability ones 88% of the time.

We tested calibration across every probability bucket. When the model says there’s an 80% chance, the actual hit rate is approximately 90%. When it says 20%, the hit rate is approximately 10%. The probabilities are trustworthy at face value.

This layer alone is better than gut feel. But it has a ceiling. It counts mentions without understanding them. A keyword with a six-quarter hit streak gets a high probability regardless of whether those mentions were enthusiastic or dismissive.

Layer 2: The LLM Context Classifier

This is where we’ve improved the model after our miss on the Dominos call.

Any two-bit high school math student can probably tell you just counting the frequency of an event does not accurately showcase the context of such an event.

To account for this every keyword/strike in the last four to eight quarters of transcripts, we extract the surrounding passage — roughly 200 words of context — and send it to an LLM. The LLM reads the actual language and classifies each mention into one of four categories:

Structural — the keyword is part of the company’s core narrative. Management brings it up unprompted as a strategic priority. These mentions persist because they’re embedded in how the company explains itself. If a strike is assigned structural they’re considered a very high conviction YES.

Promotional — management is actively pushing a product name or initiative. They’re selling the story. These mentions persist as long as the product is new and growing, but they have a shelf life. In hindsight, Craveable would’ve been promotional and if we had our new model at the time our confidence score would’ve been adjusted accordingly.

Reactive — management is responding to an analyst question or external event. They didn’t choose to say this word. These mentions are fragile — once the catalyst passes, they disappear. A good example of reactive would be 'SNAP - Family Center’

Legacy — a backward-looking reference. The keyword appears while management discusses transitions or history. The word is in the transcript, but the conversation has moved on.

Each mention also receives an intensity score from 1 to 3. A passing reference scores 1. A full paragraph scores 2. A major topic that dominates a section of the call scores 3.

The distribution of these categories — and how that distribution shifts over time — reveals whether a keyword’s frequency is sustainable or an illusion.

The Killer App: Outage

This strike is the best demonstration of why the context layer exists.

The frequency model said 85%. CrowdStrike had mentioned “outage” in six consecutive quarters following the July 2024 incident. The base rate was high, the streak was long, and the recency score was strong. Any trader looking at the numbers alone would price this above 80¢.

The context classifier told a different story. Across 12 classified mentions in the four most recent quarters, 92% were reactive and 8% were legacy. Zero structural. Zero promotional. Average intensity was 1.0 — the minimum possible. Every single mention was an analyst asking a question and management giving a one-sentence answer. Nobody was choosing to talk about the outage. They were being asked about it, and the answers were getting shorter.

The volume trend confirmed the fade: four mentions in Q1, five in Q2, down to just one in Q3.

The context-adjusted fair value was 45¢. The market had already priced it at similar levels. Outage did not hit on the March 3 call. The frequency model would have lost money buying YES at any price above 45¢. The context layer prevented that mistake.

Performance by Conviction Tier

The model’s accuracy correlated directly with conviction level:

Locks above 90¢ — two trades, both correct. Falcon Flex and Consolidation were identified as near-certainties based on structural keyword profiles. Both hit.

Strong YES between 60-89¢ — four trades, all correct. AWS, Hyperscaler, Nvidia, and Acquisition were identified as likely mentions with varying degrees of promotional vs. structural support. All hit.

Speculative zone between 30-50¢ — three strikes in this range (Microsoft, Shadow, Outage). Shadow and Outage were correctly called. Microsoft was the miss.

Strong NO below 30¢ — four trades, three correct. China, Generative AI, Signal, and Hack. Signal was the miss.

The pattern is clear: when frequency and context agree, the model is nearly perfect. When they diverge or when the probability falls into the speculative middle, variance increases. The takeaway for trading is straightforward — size your bets by conviction tier.

What We Got Wrong and Why

Microsoft was priced at 30¢ in our model. It hit. The context classifier correctly identified that Microsoft mentions had been declining for quarters — from 5 mentions per call down to zero for two consecutive quarters, with 100% promotional classification. The competitive win narrative against Microsoft had run its course.

What neither layer could see was that on February 18, CrowdStrike and Microsoft announced a major Azure Marketplace partnership. This injected a completely new narrative — not competition, but collaboration — that management was eager to discuss on the call. Our model saw a dead keyword. Reality saw a resurrected one.

If we would’ve published a CrowdStrike call analysis based on the new model, we would’ve taken into consideration this partnership and would’ve advised ‘no-action.’

Signal was priced at 20¢. It hit. The model saw scattered mentions with no pattern across eight quarters. What it couldn’t know was that CrowdStrike’s $740 million acquisition of SGNL in January 2026 turned “signal” from a generic word into a product-specific term embedded in their identity security pitch.

Both misses share the same root cause: discrete corporate events that fundamentally change a keyword’s relationship to the business. These are not predictable from historical transcripts. They require a real-time news and event monitoring layer, which is we would’ve advised ‘no-action’ or a potential lean towards yes.

Complete Technical Architecture

Training Data Construction

The frequency model was trained on a dataset of 3,144 historical quarter-keyword outcomes. We began by collecting earnings call transcripts for seven companies (CRWD, GOOGL, META, AMD, COST, TSLA, GS) through a transcript API, caching results locally to avoid redundant calls.

For each company, we defined a basket of keywords matching actual Kalshi contract strikes. The full basket covered 102 unique ticker-keyword pairs across 93 distinct keywords. We filtered to 75 pairs in the 5-95% historical hit rate range — pairs outside that range provide no variance for the model to learn from. A keyword that appears 100% of the time or 0% of the time teaches the classifier nothing.

Each row in the training set represents one historical quarter for one keyword. The label is binary: did the keyword appear on the call (1) or not (0). All features are computed using only data available before the target quarter — no future leakage.

The 43 Features

The feature extractor computes features across nine groups:

Frequency features captures raw and windowed hit rates: overall hit rate, hits in the last 2, 4, and 8 quarters, and the rate over those windows. These are the baseline signals that any manual analysis approximates by eye.

Recency-weighted features apply exponential decay to historical outcomes. We compute three variants with decay factors of 0.75, 0.85, and 0.9, plus a piecewise weighting that gives the most recent quarter 4x the weight of quarters older than 8. The top-performing single feature in the model is the 0.75 exponential recency weight, which correctly captures rapid regime changes.

Streak features track whether the keyword is currently on a hit or miss streak, the length of that streak, and a binary flag for whether the most recent call included the keyword.

Trend features measure the slope of a rolling hit rate, whether the trend is monotonically increasing or decreasing, and the ratio of recent to historical mention density.

Call-level features include average call word count, mention density per 1,000 words when the keyword is present, the standard deviation of mention counts, and the maximum mentions observed in any single call. These capture whether a keyword is a passing reference or a major discussion topic.

Mention count features track the average and maximum mention counts when present, the coefficient of variation (how volatile the count is), and the correlation between call length and mention presence.

Speaker features were designed to capture whether mentions come from the CEO, CFO, or other executives, but the transcript formatting did not support reliable speaker attribution. These features remained at zero and were excluded from the model. This feature is largely dependent upon how the transcripts are published, and if they include the title of the speaker. We are actively working on improving this feature to add to the model.

After removing zero-variance features (8 sector/speaker columns) and filling 819 null values in the length-mention correlation feature, the final feature set contained 43 usable features.

LLM Context Classification

For each keyword mention in the transcript, we extract a 600-character window of surrounding text. This passage is sent to Claude’s API (Sonnet) with a classification prompt that asks for a category (structural, promotional, reactive, legacy) and an intensity score (1-3).

The quick scoring function scans the last 4 quarters by default, with an option to extend to 8 quarters for keywords with thin recent data. It aggregates classifications into a summary that includes category distribution percentages, average intensity, dominant category, and a narrative health assessment.

The Seven Adjustment Signals

The context-adjusted probability starts from the frequency model’s output and applies seven sequential adjustments:

Signal 1 — Recent silence. If the keyword had zero mentions in the most recent quarter, the probability is multiplied by 0.6. If zero mentions in the two most recent quarters, the probability is capped at 0.25. This was the single most impactful adjustment in backtesting.

Signal 2 — Volume trend (context-gated). We compare average mention count in the two most recent quarters against the two older quarters. If volume has declined by more than 70%, and the context shows legacy mentions above 15% or structural mentions below 20%, the probability is multiplied by 0.6. If volume declined by 40-70% under the same context conditions, multiplied by 0.8. If volume increased by 50% or more and current probability is below 70%, it is boosted by 30%. The context gate prevents false penalties on keywords like Falcon Flex, where raw mention volume declined from launch hype to steady-state but the narrative strengthened.

Signal 3 — Reactive dominance. If 70% or more of classified mentions are reactive, the probability is halved. If 50-70% are reactive, it is multiplied by 0.7. Reactive mentions depend on external catalysts that may not recur.

Signal 4 — Legacy migration. If 40% or more of mentions are legacy, multiply by 0.7. If 25-40% are legacy, multiply by 0.85. Legacy mentions indicate the keyword is being discussed in the past tense.

Signal 5 — Structural boost. If 70% or more of mentions are structural and average intensity is 1.5 or above, multiply by 1.1 (capped at 0.99). Structurally embedded keywords have the highest persistence.

Signal 6 — Thin data penalty. If total mentions across all scanned quarters are 2 or fewer and the probability exceeds 40%, multiply by 0.8. Small sample sizes inflate confidence.

Signal 7 — Intensity fade. If average intensity is 1.0 or below (minimum — passing mentions only) and the probability exceeds 50%, multiply by 0.8. Low-intensity mentions signal a keyword hanging on by a thread.

Strike-by-Strike Context Classifications

Falcon Flex — 20 mentions across 4 quarters. Category split: 45% structural, 55% promotional. Intensity: 2.1 (highest on the board). Quarter-by-quarter: Q4 FY2025 was 60% structural, shifting to 60% promotional by Q2 FY2026, then back to 60% structural in Q3 FY2026. This promotional-to-structural migration pattern indicates a product transitioning from launch narrative to permanent vocabulary. No adjustment applied — frequency model’s 96.9% stands. Fair value: 95¢.

Consolidation — 17 mentions across 4 quarters. Category: 88% structural, 12% reactive, zero legacy. All four quarters showed structural mentions. Q2 FY2026 was 100% structural across all 5 classified passages. Structural boost applied: 96.4% → 99%. Fair value: 93¢.

AWS — 11 mentions across 3 of 4 quarters (silent in Q2). Category: 64% promotional, 18% structural, 18% reactive. Most recent quarter was 100% promotional at intensity 2 — management actively namedropping the partnership. One silent quarter prevents a boost but the promotional energy is strong. No net adjustment. Fair value: 75¢.

Hyperscaler — 8 mentions across all 4 quarters. Category: 50% structural, 25% promotional, 12% legacy, 12% reactive. The structural foundation represents real revenue from hyperscaler customers. One legacy mention in Q2 didn’t repeat. No net adjustment. Fair value: 73¢.

Nvidia — 7 mentions across 3 of 4 quarters. Category: 100% promotional, intensity 1.9. CrowdStrike promoting the Nvidia AI security partnership. No structural anchor means this could vanish if management finds a new partnership to highlight, but recent momentum is strong. No net adjustment. Fair value: 68¢.

Acquisition — 14 mentions across 3 of 4 quarters (silent in Q1). Category: 29% structural, 29% promotional, 7% reactive, 36% legacy — the highest legacy rate on the board. Legacy penalty applied: 80.2% → 68.2%. The keyword is transitioning from active narrative to historical references. Fair value: 60¢.

Outage — 12 mentions across all 4 quarters. Category: 92% reactive, 8% legacy, zero structural, zero promotional. Intensity: 1.0 (minimum). Volume declining from 4-5 mentions to 1. Reactive penalty and intensity fade applied: 84.8% → 42.4%. This was the model’s signature call — the frequency model was dangerously wrong, and the context layer corrected it. Fair value: 45¢.

Shadow — 2 mentions in 2 most recent quarters (emerging). Category: 50% structural, 50% promotional, intensity 2.0. New product or capability entering the vocabulary. Too thin for confidence in either direction. No adjustment from thin base. Fair value: 40¢.

Generative AI — 6 mentions across 3 of 4 quarters, with the most recent quarter silent. Category: 50% structural, 17% promotional, 33% reactive. The silence penalty fired: 51.8% → 31.1%. Management absorbed the concept into “agentic AI” without using the specific Kalshi strike phrase. Fair value: 35¢.

Microsoft — 2 mentions across 2 of 4 quarters, with 2 consecutive quarters of silence. Category: 100% promotional, zero structural. Double silence penalty applied: 57.5% → 15%. The competitive win narrative had clearly run its course by Q2 FY2026. Fair value: 30¢.

China — 2 mentions across 2 of 4 quarters. Category: 100% structural, intensity 2.0. Both mentions related to nation-state threat landscape. Structural quality is high but frequency is roughly annual. Small structural boost partially offset by thin data. Fair value: 28¢.

Signal — 1 mention across 4 quarters, with the most recent quarter silent. Category: 100% promotional, intensity 1.0. Silence penalty and thin data penalty applied: 30.8% → 18.5%. Scattered appearances with no clustering or trend. Fair value: 20¢.

Hack — 1 mention across 4 quarters (only Q3 FY2026). Category: 100% structural, intensity 2.0. CrowdStrike is a cybersecurity company but actively avoids the word “hack” in favor of sanitized terms like “threat actor” and “adversary.” 48 quarters of history show near-zero usage. Thin data keeps it flat. Fair value: 12¢.

First Strike Research publishes data-driven analysis of prediction markets, short equity research, and investigative financial journalism.

Not investment advice. Trade at your own risk.

Our analysis examines why keyword frequency models — including systematic quantitative approaches — may be dramatically underestimating this particular contract.

The Surface-Level Pattern

First Strike Research maintains API access to every single public-facing company earnings call transcript, and our machine learning system poured over the data:

Across Domino’s earnings calls from 2018–2024, “craveable” received zero mentions — a complete absence spanning at least six years of quarterly calls.

The phrase then appeared for the first time on the Q3 2025 call in October 2025.

A basic recency model, seeing a single-quarter appearance after years of zeros, interprets this as a marginal or one-off usage — something to discount rather than extrapolate.

Current Kalshi YES pricing sits around 37¢, implying the market believes “craveable” is unlikely to appear on the upcoming Q4 call. We believe this to be false — and the magnitude of the mispricing is the largest we have identified in our keyword contract research to date.

Why This Matters Beyond One Trade

Our model, trained on historical mention frequencies, cannot easily detect when a word transitions from “absent from corporate vocabulary” to “the company’s new defining brand adjective.” The model notes a single appearance after years of nothing and assumes mean reversion back toward zero.

So in turn we layer qualitative analysis — event timing, social sentiment, brand strategy context, and management communication patterns — on top of our statistical framework.

First Strike’s proprietary machine learning system has the probability for “Craveable” being said at:

68.4% — YES

Bear in mind — the above is solely for raw transcript data (no social analysis).

A 31 percentage point over-discount applied to the contract on the numbers alone. When we drill into the brand context and social sentiment, we’ve found the true odds of Domino’s saying “craveable” even higher. This is based on the structural brand tailwinds and business trends we’ve found through our research.

This deep research is provided below to our paid subscribers.

Subscribe now

DISCLOSURE:

The author of this article holds the following position:

“CRAVEABLE” — YES

Read more

It was a volatile market based in part on relatively low liquidity. We had a final dollar cost average of ~50 cents a “NO” contract. As we accurately predicted, Snapchat Family Center was NOT mentioned on SNAP’s earning call.

A couple considerations on why:

• The news-cycle came and went, SNAP previously announced their improvements to the Family Center on the call prior.

• The improvements were primarily focused around legal mitigations after a lawsuit. This is generally an ugly topic, and a topic management doesn’t want to bring back up once it’s been put to bed.

• SNAP was more focused on pushing the growth narrative.

• After a strong earnings beat, analysts are even less likely to waste their one question to ask about a lawsuit that has already been settled out and “priced-in” to the equity shares. If Family Center wasn’t mentioned in the prepared remarks the odds of them saying it in the Q&A are further reduced (virtually to zero.)

• This presents an appealing live trade opportunity which is why you should be dialed into all earning calls you have a position on.

  ---

Subscribe now

Looking forward at company earnings mention contracts

A pattern we’ve noticed is if a company has a top/bottom line miss, they are far more likely to heavily stress speaking about financial metrics in their prepared remarks. Even more so if the stock is aggressively selling off.

We believe this adds more weight to “NO” contracts around ‘fluff’ some recent examples:

• Cane Sugar “NO” - not said on the Coke call after a sales miss.

• Retro “NO” - also not said on the Coke call after a sales miss.

Given that earnings reports release roughly 30-60 minutes before the call this is basically a live trade. We will not be able to bake in this metric to our machine learning model. To avoid spamming your inbox before every earnings call if we spot something we’ll post it to either Substack Notes or our Twitter / X.

We’ve made improvements to our machine learning system

Our next market write up will include the new weights we’ve added to our machine learning algo. We’ve added overall call length to assign more weight to the YES side of trades taking into account longly winded CEOs like Elon Musk who are apt to ramble and subsequently hit more Ys

Limit Orders are Your Friend

Are you trading? Or are you gambling? You should strongly consider setting limit sells at a comfortable price target. Buying a 20c contract and selling back at 40c is a solid trade and represents a 100% percent ROI. Dumb money hits market orders, and rides out until resolution. Real traders close out their positions with a healthy ROI and avoid resolution risk. You can either trade outcomes, or probability flow. While you can be profitable in both, trading the waves of probability will make you more money in the long run.

Also be sure to check out Bernard B’s:

Bernard B.

Prediction Market Dictionary

This page is an attempt to catalog and define the rapidly evolving vernacular used in prediction markets such as Kalshi and Polymarket. Readers on desktop can use the table of contents on the left for ease of navigation…

Read more

6 months ago · 6 likes · 3 comments · Bernard B.

Up next

By the end of the week be prepared for another Kalshi earnings mention contract our machine learning model shows to be mispriced. Our next write up will include the added calculation of average call runtime. Paid subscribers will get access to additional social sentiment and weight added to the calculation based on the questions we predict wall street analysts will ask on the call.

Kalshi traders currently have the odds of “Family Center” being said on the upcoming Snap Inc. earnings call at an 87% chance. We believe these odds are drastically overstated. We utilized our machine learning system and social sentiment analysis to provide First Strike readers more accurate odds for this market.

The market is here

Note: The earnings call is TOMORROW, Kalshi added this market almost last minute.

Disclaimer: This is research supporting a financial derivative, and governed by our terms of service.

The pattern

First Strike Research maintains API access to every single public-facing company’s earnings call transcripts, and our machine learning system pored over the data:

• Across 35 Snap earnings calls (2014–2025), “family center” received zero mentions for approximately 10 quarters after the product launched in mid-2022

• The phrase then appeared in one call—the FY2024 Q4 call on February 4, 2025

• Current Kalshi YES pricing sits around 87¢, implying an extraordinarily high probability that “family center” has become a fixture of Snap’s quarterly script

We believe this to be false (at least going forward).

The numbers

First Strike’s proprietary machine learning system has the probability for “Family Center” being said at:

9.2% — YES

Obviously, our model which is trained on historical mention frequencies cannot easily detect when a talking point has transitioned from “product launch announcement” to “quietly embedded feature that no longer warrants airtime.” The model notes a recent mention and extrapolates continued discussion.

Bear in mind—the above is solely for raw transcript data (no social analysis).

9.2% YES which equates to a roughly 78-point overconfidence premium applied to the contract on the numbers alone. When we drill into the social sentiment and regulatory context, we’ve found the street-adjusted odds sit around 15–20%—still dramatically below the market’s 87%.

This is based on analyst silence, peer company behavior, management communication patterns, and event sequencing we’ve found through our research.

Below, paid subscribers get the full breakdown why the January announcement is a pricing trap, what Snap's Q3 call and Meta's Q4 call reveal about management's actual language habits, where the analyst community is focused (hint: not here), and our complete risk scenario analysis. On the fence? You can read about our previous success here:

DISCLOSURE:

The author of this article holds the following position:

SNAP MENTION “FAMILY CENTER” — NO

THIS ARTICLE WAS ORIGINALLY PAY WALLED FROM HERE DOWN - SINCE THE EVENT HAS CONCLUDED WE HAVE REMOVED THE PAYWALL.

Subscribe now

Subscribe, so you don’t miss market intelligence like this.

Paid subscribers, thank you for your patronage. I care about maintaining a strong value proposition. If you have any questions, comments, concerns, or ideas for improvement, please let me know!

Historical Silence

To understand why this keyword is overpriced, it helps to examine what “family center” actually is and why its mention history is so remarkably thin.

Family Center is Snapchat’s in-app parental tools suite, launched in mid-2022. It allows parents to see who their teen is communicating with, set content controls, and monitor usage. Despite existing for over three years and spanning 14+ quarterly earnings calls, management has mentioned it exactly once.

That single mention came on the FY2024 Q4 call (February 4, 2025): “We also announced new location sharing features in Family Center...” — a passing reference nested within a product update, not a standalone talking point.

A 2.9% base rate across 35 calls is not a pattern that supports 87% pricing. It’s a pattern that supports single-digit probability, with room for qualitative adjustment.

The January Catalyst

The Kalshi market’s 87% price almost certainly stems from two events in late January 2026:

January 20, 2026: Snap settled the K.G.M. addiction lawsuit, avoiding trial and CEO Evan Spiegel’s testimony. The settlement terms were minimal—Snap’s public statement said only that they were “pleased to resolve amicably.”

January 22, 2026: Two days later, Snap announced the most significant expansion of Family Center in its history—new time-spent insights for parents, friend trust signals, and AI-powered parental controls. The announcement received major press coverage from The Hill, TechCrunch, Mashable, and IT Brief UK.

For retail Kalshi traders, the logic is obvious: Big product announcement 13 days before earnings = definitely gets mentioned on the call.

But this logic doesn’t survive contact with the data.

The Q3 2025 Precedent

The cleanest comparable is the Q3 2025 earnings call from November 5, 2025—just three months ago.

During that call, Snap faced active regulatory pressure. Australia had just passed its social media age verification law. Snap discussed regulatory impacts and age verification on the call.

“Family center” was mentioned zero times.

Management discussed safety. They discussed regulatory compliance. They discussed parental tools in generic terms. But they did not use the specific phrase “family center”—which is the only thing that matters for contract resolution.

This is the single most important data point in the analysis. When Snap had every reason to mention Family Center by name on the Q3 call—active regulation, safety discussion, product relevance—they chose not to. The contract resolves on the exact phrase, and Snap’s management defaults to discussing safety without naming the specific product.

Peer Confirmation: Meta’s Silence

Meta reported Q4 2025 earnings on January 28, 2026—six days before the Snap call. Meta owns “Teen Accounts,” a directly comparable parental controls feature.

Despite facing the same regulatory environment, the same Congressional pressure on child safety, and the same news cycle around teen social media usage, Meta made zero mentions of parental controls or teen safety in either prepared remarks or Q&A.

This suggests a sector-wide pattern: social media companies are not volunteering safety feature discussions on earnings calls unless directly forced by analyst questions. The earnings call is for financial results—revenue, DAUs, margins, guidance. Product safety features are PR and blog post material.

The Analyst Silence

We reviewed 8+ sell-side analyst previews for the Snap Q4 2025 call across Yahoo Finance, Zacks, MarketBeat, and other sources.

Zero previews mention Family Center, teen safety, parental controls, or child safety regulation as an expected topic for the call.

The analyst community is focused on:

• Snapchat+ subscriber trajectory and monetization

• The Perplexity AI partnership and its revenue contribution

• North America DAU growth and engagement trends

• 2026 guidance and margin trajectory

If the street thought “family center” was an 87% probability topic, at least one analyst preview would flag it. None did.

The Retail Anchoring Problem

The 87¢ YES price exhibits classic retail anchoring bias combined with thin-market inefficiency.

The Kalshi spread on this contract is 25¢ (YES at 87¢, NO at 38¢). A 25-cent spread signals extremely thin liquidity. In thin markets, a handful of uninformed retail traders can push the price far from fair value.

The likely chain of reasoning for YES buyers:

1. See headline: “Snap Expands Family Center with AI Parental Controls”

2. Think: “They just announced this—of course they’ll mention it on earnings”

3. Buy YES at 87¢

4. Never check: historical base rate (2.9%), Q3 precedent (zero mentions despite safety discussion), or whether product announcements typically translate to earnings call mentions (they often don’t)

We searched the Kalshi subreddit, StockTwits, prediction market forums, and fintwit. Zero discussion of this specific contract. The price appears entirely uninformed.

The Lawsuit Settlement Dynamic

The K.G.M. lawsuit settlement on January 20 creates a nuanced risk. On one hand, the settlement + Family Center announcement timing (2 days apart) suggests coordinated PR—Snap wanted to pair “we settled” with “here’s what we’re doing proactively.”

On the other hand, Snap’s settlement statement was deliberately minimal. Management typically does not volunteer lawsuit discussion on earnings calls unless the financial impact is material. Even if asked about safety, Snap can—and historically does—discuss “parental tools,” “age verification,” and “trust and safety investments” without using the specific phrase “family center.”

The Q3 call is proof: safety discussion occurred without the keyword. The contract doesn’t care about intent or topic. It cares about two specific words spoken together.

Forward Probability Assessment

Four factors push the realistic forward probability well below the Kalshi market price:

Historical base rate: 2.9% across 35 calls. The product has existed for 3.5 years and been mentioned once. This is not a fixture of Snap’s earnings script.

Q3 2025 precedent: Safety and regulatory discussion occurred without the phrase. Management defaults to generic safety language on earnings calls.

Analyst silence: Zero sell-side previews flag Family Center as an expected topic. The street is focused on advertising, DAUs, and guidance.

Peer behavior: Meta’s Q4 call omitted teen safety discussion entirely despite owning comparable features and facing identical regulatory pressure.

After adjusting for the January 22 announcement catalyst (which provides a modest probability lift), a reasonable forward probability for “family center” on the Q4 2025 earnings call sits in the 15–20% range.

Analytical Implications

At current NO pricing around 38¢, the implied market probability is approximately 87%. If the adjusted probability sits closer to 18%, this represents potential edge of approximately 67–70 percentage points.

At our model’s 9.2% estimate, the edge is even wider—though we believe the January announcement timing justifies a modest qualitative lift above the pure statistical output.

Expected value calculation at 18% fair value:

• NO costs 38¢, pays $1.00 if “family center” is not said

• EV = (0.82 × $0.62) − (0.18 × $0.38) = +$0.44 per contract (+115% ROI)

Several considerations warrant caution:

Q&A tail risk: Analyst questions are inherently unpredictable. A single inquiry about the lawsuit settlement could prompt management to pivot to Family Center as a defense. We estimate this at 5–8% probability.

Prepared remarks wildcard: If the IR team scripted a safety paragraph citing Family Center by name, the trade loses. This is the biggest unknown, estimated at 10–12% probability. Q3 precedent suggests they don’t script it, but it cannot be ruled out.

Informed flow: If the 87¢ price reflects someone with inside knowledge of prepared remarks (unlikely but possible), the analysis is wrong. Monitor for unusually large YES orders before the call.

Traders evaluating this thesis should size accordingly and recognize that even well-researched positions carry meaningful uncertainty.

Key Monitoring Points Before the Call

• Sell-side notes published February 3–4: If any analyst suddenly flags Family Center, probability jumps to 30–35%

• Unusual Kalshi order flow: Large YES buys ($10K+ single fills) could signal informed trading

• Snap IR communications: Any pre-call press release or investor letter mentioning Family Center would be a warning sign

If none of these appear, our NO trade thesis remains intact.

And—the simplest way to understand CrowdStrike’s competitive vulnerability requires absolutely no technical background. One of their biggest competitors: Microsoft has a cybersecurity product (Defender) for Endpoint built into their enterprise license suite at no additional cost. Approximately 60% of CrowdStrike’s Fortune 500 customer base already holds E5 licenses. So these organizations are, in practical terms, paying CrowdStrike a premium subscription for endpoint protection they already own through their existing Microsoft agreement.

By way of simple analogy this is like purchasing a home that comes with a free professionally monitored security system (Microsoft.) Only to then turn around and pay a different company (CrowdStrike) to install a second system on top of it. This is a redundancy that was once justified only by the market’s perception that Falcon was the categorically superior product.

Our research below will show CrowdStrike's reported 97% retention rate is structurally misleading and most outage-affected customers haven't reached renewal yet. We present evidence of a joint SEC/DOJ investigation, $500M+ in litigation surviving dismissal, and a contract renewal cliff the market has not priced.

In a follow-up article we will be presenting Open Source Intelligence (OSINT) collected by First Strike investigators which reveals CrowdStrike’s arrogant “cybersecurity Jesus” complex headlined by their social prestige and backstopped by a culture of catastrophic QA, pushy sales tactics, and a befuddled customer support network which is often regarded as painstakingly slow.

Don't miss out, subscribe now!

CrowdStrike trades at 25.8x revenue with a 95x forward P/E we believe the market hasn't yet seen what happens when the contracts come due. On July 19, 2024, a single faulty content update crashed 8.5 million Windows systems in the largest IT outage in history. Airlines grounded 16,896 flights. Hospitals went dark. Fortune 500 companies absorbed $5.4 billion in losses. But because CrowdStrike locks enterprise customers into 1-3 year prepaid contracts with no termination for convenience, and the churn from that catastrophe has just begun to surface in reported financials. The first major renewal wave starts now. Meanwhile, the SEC and DOJ are jointly investigating the company's revenue recognition practices — a dual-agency probe that, outside of FCPA cases, is rare and typically signals potential criminal exposure.

At the helm is CEO George Kurtz, who is the only cybersecurity executive in history to preside over two global IT outages caused by faulty updates. He declined to testify before Congress about the incident; opting to send Adam Meyers, the company’s Senior Vice President of Counter Adversary Operation instead.

First Strike investigators also uncovered dozens of testimonies from former employees, customers, and adjacent cyber security professionals regarding the arrogant and reckless culture that is plaguingthe company.

• Premium valuation disconnected from decelerating fundamentals. CRWD trades at 25.8x P/S and 95.2x forward P/E — more than double peer averages — while revenue growth has decelerated from 36% to 22%, ARR growth has fallen from 35% to 23%, and net revenue retention has dropped to a record-low 112%. Free cash flow margin compressed from 31% to 27%, moving further from management’s 34-38% target.

• The July 2024 outage was the largest IT failure in history — and the financial impact is structurally delayed. A faulty update, negligently pushed by CrowdStrike caused $5.4 billion in estimated Fortune 500 losses and over $10 billion in total global economic damage. CrowdStrike’s 1-3 year contracts stipulate no termination for convenience which means tranches of customers have not yet reached renewal.

• Competitors are actively converting pipeline from the outage. Palo Alto Networks confirmed active discussions with companies seeking to switch providers. SentinelOne also reported record pipeline and a record number of competitive wins, while growing revenue at 32% way faster than CrowdStrike.

CrowdStrike’s market dominance hides cracks in the foundation

CrowdStrike commands an impressive 21-22% market share in endpoint protection, making it the category leader ahead of McAfee and Microsoft Defender.

The flagship Falcon platform is deployed across 60% of the Fortune 500 and 70% of the Fortune 100, this has afforded CrowdStrike a deep competitive moat through its cloud-native architecture and real-time threat intelligence. The company reached a $118 billion market capitalization and joined the S&P 500 in 2024—the fastest any pure-play cybersecurity firm has achieved that milestone.

However, we believe this growth is decelerating rapidly. Annual recurring revenue growth has fallen from 35% in Q3 FY2024 to 23% in Q4 FY2025, while revenue growth declined from 36% to 22% over the same period.

In our opinion, one of the most concerning problems with CrowdStrike is their net revenue retention rate. By way of context, NRR is one of the most important KPIs in SaaS as it shows whether a company’s customer base is growing more or less valuable overtime. This important metric demonstrates CrowdStrike’s ability (or lack-there-of) to upsell, cross sell, and avoid churn.

NRR at CrowdStrike—dropped to 112%, the lowest in company history, down from 119% just four quarters earlier. Net new ARR of $153 million in Q3 FY2025 represented a significant decline from the $223 million record set the prior year.

The valuation premium appears increasingly disconnected from these fundamentals. CrowdStrike trades at a forward P/E of 95.2x versus an industry average of 29.5x, and a P/S ratio 2.2 times higher than peer companies including Palo Alto Networks, SentinelOne, and Zscaler. Free cash flow margin compressed from 31% in FY2024 to 27% in FY2025, moving further from management’s stated target of 34-38%. GAAP operating margin swung from near-breakeven to negative 3%, generating a $120.4 million operating loss in FY2025 compared to prior-year profitability.

Competitive threats are intensifying from multiple directions. Microsoft’s Defender offering has captured 12.7% market share through aggressive bundling with E5 enterprise licenses—meaning many CrowdStrike customers already have unused Microsoft security capabilities they could consolidate to at no incremental cost. By way of simple analogy without technical nuance: this is like buying a house which includes a free security system (Microsoft) only to turn around and purchase a completely separate system.

Palo Alto Networks’ $25 billion CyberArk acquisition and “platformization” strategy directly targets CrowdStrike’s enterprise customer base.

SentinelOne is showing impressive growth at 32-33% annually faster than CrowdStrike all while achieving its first-ever profitable quarter while offering similar capabilities at lower valuations.

The July 2024 outage was unprecedented in scale and cause

At 04:09 UTC on July 19, 2024, CrowdStrike released a faulty Falcon sensor content update that crashed 8.5 million Microsoft Windows systems worldwide—the largest IT outage in the history of information technology.

The root cause was simple: CrowdStrike pushed an update with an additional data field that the Falcon sensor was not built to handle. By way of simple analogy CrowdStrike put diesel fuel in a gasoline car. This in turn crashed windows machine into endless boot loops.

To get technical: Channel File 291 provided 21 input fields when the Falcon sensor expected only 20, causing an out-of-bounds memory read that triggered Windows Blue Screen of Death errors and boot loops.

The Falcon sensor operates at the Windows kernel level with the highest system privileges. When it crashed, it took down entire operating systems. BitLocker-encrypted machines required unique 48-digit recovery keys for restoration, and all affected systems needed manual intervention—technicians booting into Safe Mode to delete a single faulty .sys file, one machine at a time.

The damage spread across every sector of the global economy. Airlines cancelled 16,896 flights within 72 hours.

Parametrix Insurance estimated $5.4 billion in Fortune 500 losses alone, with total global economic damage exceeding $10 billion. Delta Air Lines was hit hardest of all, cancelling 7,000+ flights, stranding 1.3 million passengers, and suffering $550 million in total losses ($380 million revenue, $170 million expenses).

CrowdStrike’s crisis response drew sharp criticism. CEO George Kurtz’s initial Twitter statement identified the problem but notably omitted any apology.

Their contract structure creates a delayed churn trap

The most critical element of our thesis may be what current financials cannot yet show. CrowdStrike requires upfront payment for the entire subscription term—typically 1-3 years—and explicitly does not allow termination for convenience. This contractual structure creates a significant lag between customer dissatisfaction and visible churn.

Enterprise customers locked into multi-year agreements signed before July 2024 cannot switch providers until their contracts expire. A three-year deal signed in Q3 2023 won’t reach renewal until Q3 2026—more than two years after the incident. Even one-year contracts signed in early 2024 are only now approaching their first renewal window. The Q1-Q3 2025 period represents the first major wave of post-incident contract renewals, meaning the true test of customer loyalty is only beginning as of February 2025.

SentinelOne’s CEO observed that “companies do not make snap decisions and need to figure out how to make the transition. Sales cycles typically last 9 to 12 months.”

Competitors report active customer interest. Palo Alto Networks’ CEO confirmed “a number of discussions with companies looking to change vendors,” while SentinelOne reported a “record pipeline” and “significant pipeline pickup” following the outage, later citing “record number of wins against closest competitor.” CrowdStrike acknowledged losing accounts in the “managed security services space” and has deployed “Customer Commitment Packages”—discounts, free products, and extended terms—that reduced quarterly revenue by $10-15 million while generating “longer-term deals” designed to lock customers in past the immediate churn risk window.

CrowdStrike reports a 97% gross retention rate, but this metric only measures customers who reached renewal—it cannot capture churn intent among the majority still locked into multi-year contracts. The true retention picture will only emerge over the next 18-24 months as the contract renewal wave progresses.

Legal and regulatory exposure extends beyond Delta

Delta Air Lines filed suit against CrowdStrike on October 25, 2024, in Fulton County Superior Court, Georgia (Case No. 24CV013621), seeking $500 million or more in damages plus punitive damages and legal fees. The complaint alleges gross negligence, breach of contract, computer trespass under Georgia law, and willful misconduct, claiming CrowdStrike “forced” untested updates through an “unauthorized door” within Microsoft’s operating system. Delta’s attorney David Boies argued CrowdStrike’s initial remediation assistance “simply referred Delta to CrowdStrike’s publicly available remediation website.”

CrowdStrike filed a countersuit the same day in federal court (N.D. Georgia, Case No. 24-cv-04904), seeking declaratory judgment that liability is capped at contract limits—which external counsel characterized as “single-digit millions”—and blaming Delta’s “antiquated IT infrastructure” for the extended recovery period. In May 2025, Judge Kelly Lee Ellerbe ruled that Delta’s claims may proceed, allowing gross negligence and computer trespass claims to advance while dismissing most fraud allegations. If Delta proves gross negligence or willful misconduct, standard contractual liability caps may be invalidated, exposing CrowdStrike to the full claimed damages.

The most significant legal development came in June 2025, when CrowdStrike disclosed an ongoing SEC and DOJ joint investigation into revenue recognition practices. The probe focuses on a $32 million Carahsoft-IRS transaction from 2023, structured as four $8 million payments with reportedly no products ever purchased. CrowdStrike later excluded approximately $26 million from ARR citing “distributor-transferability issues.” Investigators are interviewing former staff about potential “pre-booking” or “channel-stuffing” practices and reviewing Sarbanes-Oxley compliance questionnaires. Dual SEC/DOJ involvement outside FCPA cases is unusual and suggests potential criminal exposure. Disclosure Insight research flagged elevated restatement risk.

Additional proceedings include a dismissed passenger class action against CrowdStrike (now on appeal to the 5th Circuit) and an active passenger class action against Delta where some claims survived dismissal in May 2025. Insurers who paid first-party claims to affected customers are expected to pursue subrogation claims. Total insured losses from the incident are estimated at $400 million to $1.5 billion.

George Kurtz’s track record raises governance questions

CEO George Kurtz co-founded CrowdStrike in February 2012 after serving as McAfee’s Executive Vice President and Worldwide Chief Technology Officer. His tenure at McAfee included oversight of a strikingly similar incident: On April 21, 2010, McAfee released a faulty antivirus update that mistakenly flagged the critical Windows XP file “svchost.exe” as malicious, sending systems into endless reboot loops. Internal documents cited by ZDNet blamed “inadequate” quality control. McAfee was acquired by Intel for $7.7 billion shortly after. Kurtz resigned in October 2011.

Kurtz is now the only cybersecurity CEO to have been at the center of two global IT outages caused by faulty security software updates—a pattern that raises fundamental questions about quality control culture under his leadership.

Following the July 2024 incident, the House Homeland Security Committee formally demanded that Kurtz personally testify before Congress. CrowdStrike declined, stating Kurtz was “not the appropriate witness,” and instead sent SVP Adam Meyers to the September 24, 2024 hearing. Industry observers expressed surprise. Josh Aaron, CEO of Aiden Technologies, stated it was “a little shocking to me that CEO George Kurtz didn’t agree to also testify.” Meyers acknowledged during testimony that the faulty update was distributed “to all customers in one session” and that the validator that failed had been “in place for over a decade.”

Kurtz’s compensation totaled approximately $47 million in FY2024 and $35.2 million in FY2025—158 times median employee pay. The FY2024 package represented a 29% increase over the prior year despite the catastrophic July 2024 incident occurring mid-fiscal year.

SEC Form 4 filings document substantial insider sales in the months preceding the outage. Kurtz sold approximately $28.4 million in stock on January 12, 2024, $25.5 million on March 21, 2024, and $17 million on May 3, 2024—totaling over $70 million in the six months before the incident. The company characterizes these as routine administrative sales to cover tax withholdings on vesting RSUs under 10b5-1 plans. CrowdStrike insiders recorded 41 sales and zero purchases over a twelve-month period.

Delayed reality

Our thesis does not solely rest on speculation about whether the July 2024 outage damaged customer relationships—it rests on the structural reality that most of that damage cannot yet appear in reported metrics. The combination of 1-3 year upfront-paid contracts, 9-12 month enterprise sales cycles, and no termination for convenience means the true retention picture will only emerge through the Q1 2025 to Q3 2026 renewal window.

The company faces this reckoning with a 25.8x P/S multiple that assumes continued premium growth, despite clear evidence of deceleration: revenue growth down to 22%, ARR growth down to 23%, net revenue retention at a record-low 112%, and FCF margin compressing from 31% to 27%. Competitors are growing faster (SentinelOne at 32%), executing aggressive platform strategies (Palo Alto’s $25 billion CyberArk deal), or bundling at zero incremental cost (Microsoft Defender).

Legal exposure compounds operational risk. The Delta lawsuit seeks $500 million and survived a motion to dismiss on gross negligence grounds—the specific claim that could void CrowdStrike’s contractual liability caps. The SEC/DOJ joint investigation into revenue recognition represents potential restatement and enforcement risk. And CEO George Kurtz’s refusal to testify before Congress, combined with his unique distinction as the only cybersecurity executive to oversee two global IT outages, raises fundamental governance questions.

The bull case assumes customers will forgive and renew, the SEC investigation will resolve without material consequence, and competitive dynamics will not erode the premium valuation. The bear case observes that none of these assumptions have yet been tested—and the tests begin now.

Subscribe now

Subsequently, all contracts bought at 30 cents moved to a dollar. This represented a nearly 233% return to traders that followed our research.

Our primary confidence in this trade (which was revealed exclusively to paid subscribers, among other intelligence tranches) was news cycle based. We determined the 70 percent chance of them saying it was grossly overpriced and the clear upside was in them NOT saying it.

The “Condiment Bar” news cycle, came and went. And your average Kalshi trader who trades off vibes based their positions on “Well, if they said it three calls in a row they’ll certainly say it again.”

Don’t take our word for it

Over the past 24 hours I have been inundated on my personal line, Discord, and Substack with the overwhelming gratitude of our readers.

This past week, First Strike has seen an over 1,000% increase in free subscribers, and in turn a 75% increase in paid subscribers. To new subscribers, both paid and unpaid: Thank you!

Lately, our focus has been on event contracts. Primarily because mis-pricing is abundant and we have demonstrated that there is real edge here for those that are willing to do the research, and trade with a “data first” approach.

This weekend we will release our write up on Crowd Strike - and another Kalshi Earnings Call Mention Market write-up so stay tuned.

Both Kalshi and Polymarket had versions of this mention market. What happened next is an interesting case of regulated vs. un-regulated markets.

Kalshi, a CFTC-regulated exchange, resolved based on the rules as written.

Polymarket, an unregulated exchange changed the rules after the event was over, then resolved said event to YES.

This of course, has left Polymarket traders absolutely livid—

Share

But, how?

On Kalshi, the rules were clear from the start. “Bros” meant “Bros.” The speakers on the call said “Brothers.” The textual interpretation was straightforward, even if some traders disagreed with it. The market resolved according to the written criteria that existed when traders entered their positions.

On Polymarket, the outcome was less certain—so they simply made it certain. After the event concluded, the resolution source was amended to treat “Warner Brothers” and “Warner Bros” as equivalent. After almost three days of back-and-forth disputes—the market resolved to YES.

The CFTC Exists for This Exact Reason

I don’t want to seem like I’m overtly praising CFTC. But let’s be clear about what the CFTC’s oversight of Kalshi actually provides:

1. Immutability: Contract terms are filed with the CFTC and cannot be changed post-issuance on a whim.

2. Transparency: Resolution criteria must be documented and applied in a uniform manner.

3. Recourse: Traders have a regulatory body and the National Futures Assoc. to appeal to if an exchange behaves improperly.

4. Regulatory Accountability: Kalshi can’t just decide who wins after the fact.

Polymarket operates offshore, and outside CFTC jurisdiction. There’s no filing requirement. There’s no regulatory oversight. There’s no appeals process beyond hoping the UMA oracle or Polymarket’s team rules in your favor.

When the rules are whatever Polymarket says they are—and whenever they want to say it—it adds risk to traders.

But isn’t Polymarket Decentralized?

Yes, settlement technically runs through UMA’s oracle system. But when Polymarket itself amends the resolution criteria fed to that oracle, “decentralization” is just theater. The oracle resolves based on inputs. Polymarket controls the inputs.

Decentralized infrastructure executing centralized decisions is just centralization with extra steps.

The Precedent Problem

Polymarket’s has now set the precedence:

The rules are suggestions. We can change them if the outcome feels wrong to us.

How do you price a contract when the strike price can move after expiration?

For small retail traders, this might not matter much. For anyone trying to run a serious strategy, deploy real capital, or build systematic approaches to prediction markets, Polymarket just revealed itself as unacceptably risky not because of market risk, but because of platform risk.

This is a rare case, where regulators ended up being cheered by traders.

Leave a comment

Disclaimer

This article is for informational and educational purposes only and does not constitute investment advice, trading advice, or a recommendation to buy, sell, or hold any position in any event contract or financial instrument. Event contracts involve risk, including the potential loss of your entire investment. Past performance is not indicative of future results.

The author may hold positions in contracts discussed on various prediction market platforms. Contract interpretation and regulatory analysis is opinion only. The author is not providing Commodity Trading Advice (CTA) by publishing this article. Readers should conduct their own research and consult with a qualified financial advisor before making any trading decisions.

You can find the market on Kalshi here.

Disclaimer: This article is for educational and informational purposes only. It does not constitute investment advice, trading recommendations, or solicitation to buy or sell any financial instruments or prediction market contracts. All analysis represents the author’s research and opinions based on publicly available information. Past performance of any trading strategy does not guarantee future results. Readers should conduct their own due diligence and consult with qualified professionals before making any trading decisions. Prediction market trading involves substantial risk of loss.

Kalshi traders currently have the odds of “Condiment Bar” being said on the upcoming Starbucks earning call at a 76% chance. We believe these odds are drastically overstated. We utilized our machine learning system, and social sentiment analysis to provide First Strike readers more accurate odds for this market.

Our analysis examines why keyword frequency models—including systematic quantitative approaches—may be overestimating this particular contract.

The Surface-Level Pattern

First Strike Research maintains API access to every single public facing companies earning calls transcripts, and our machine learning system poured over the data:

• Across 48 Starbucks earnings calls (2014–2025), “condiment bar” received zero mentions for approximately a decade

• The phrase then appeared in four of the last five quarters

• A basic recency-weighted model interprets this as an upward trend

Current Kalshi YES pricing sits around 76¢, implying a high probability that “condiment bar” has become a fixture of Starbucks’ quarterly script. We believe this to be false (at-least going forward.)

Why This Matters Beyond One Trade

Obviously, our model which is trained on historical mention frequencies cannot easily detect when a talking point has transitioned from “forward-looking initiative” to “completed checkbox.” The model notes rising mentions and extrapolates continued discussion.

So in turn we layer qualitative analysis—event timing, social sentiment, and management communication patterns—on top of statistical frameworks.

First Strike’s proprietary machine learning system has the probability for “Condiment Bar” being said at:

31.6% - YES which equates to a roughly 40% over-confidence premium applied to the contract on the numbers alone. When we drill into the social sentiment we’ve found the true odds of them saying Condiment bar even lower. This is based on social head-winds, and business trends we’ve found through our research.

This deep research is provided below to our paid subscribers. If you’re not a paid subscriber become one today!

DISCLOSURE:

Read more

Based on our examination of confirmation mechanics, competitors, and structural factors, our analysis concludes that Waller’s true probability sits closer to 22%

Which is nearly double his current market price. We will also provide a bull case that Christopher Waller may be the next Fed Chair given the current socio-economic-political environment. Paid subscribers will have access to our early warning intelligence signals at the end of the article!

Current State of Play

Treasury Secretary Scott Bessent confirmed on January 21, 2026 that Trump has narrowed the field to three finalists.

These odds reflect public perception of Trump’s statements rather than genuine insider intelligence. The race experienced dramatic volatility after Trump’s January 16 comment about wanting to “keep Hassett where he is”—a 26-point sell-off.

As expected markets are reacting to public signals, not tangible “inside the White House” information.

Waller’s Profile

Christopher Waller brings a unique combination of Trump loyalty, academic credibility, and institutional knowledge.

Credentials:

• Current Fed Governor since December 2020 (term through January 2030)

• Research Director, Federal Reserve Bank of St. Louis (2009-2020)

• Ph.D. Economics, Washington State University

• 100+ peer-reviewed publications on monetary theory and central bank policy

2020 Confirmation: Waller was confirmed 48-47 with five Democratic senators crossing party lines. This is a proven track record of bipartisan ability for confirmation.

December 2025 White House Interview: President Trump conducted several in-person interviews with his short list. Waller was described by two senior officials as having a “strong interview.” It is believed his interview was predominately focused on the labor market and job creation strategies. During this interview President Trump was flanked by Treasury Sec. Bessent, Susie Wiles, and Dan Scavino his Chief of Staff and Deputy Chief of Staff, respectively.

The Case for Waller

1. Senate Math

All roads lead to a 13-11 Republican majority on the Senate Banking Committee.

However —

Senator Thom Tillis (R-NC) publicly declared on January 12 that he will oppose any Fed nominee until the DOJ’s criminal investigation into Powell is resolved. The important keyword to remember here is: ANY.

With Tillis voting no, the committee splits 12-12—blocking any nomination from advancing.

Waller has a unique advantage: Among the four finalists, only Waller has demonstrated bipartisan confirmability. If the DOJ matter involving Powell is not resolved, we expect trump to select the candidate most likely to pass when the window opens.

That candidate is Waller.

2. The Larry Summers Victory

When the Fed began aggressive rate hikes, Larry Summers argued unemployment would need to rise to 6%+ to bring inflation down—implying roughly 4 million job losses. Waller countered with Beveridge Curve analysis suggesting inflation could fall without mass layoffs.

Subsequently—Inflation dropped from 8%+ to near 2%. Unemployment remained below 4%. Waller was correct; Summers was wrong.

Trump overwhelmingly values: economic policy elites (Waller outperformed a Harvard economist) and Trump also wants to be able to take credit for a soft landing and strong job market.

3. The Policy Alignment

Throughout 2025, Waller executed a strategic dovish pivot:

• July 2025: Dissented when FOMC held rates steady, arguing for a 25 basis point cut

• November 2025: Made “The Case for Continuing Rate Cuts” in London, citing labor market concerns

What Waller posits is directly aligned with Trump’s routinely public calls for “lower interest rates”

Competitor Vulnerabilities

Kevin Warsh (51% Odds)

The Estée Lauder Problem: Warsh married Jane Lauder, granddaughter of Estée Lauder’s founder. His father-in-law Ronald Lauder is a Trump ally who planted the Greenland annexation idea. This connection provides access but creates a nepotism narrative that Senate Democrats will exploit.

Policy Inconsistency: Warsh resigned from the Fed in 2011 after opposing QE2, warning about inflation risks from monetary expansion. His current advocacy for aggressive rate cuts represents a complete reversal that confirmation hearings will expose.

The Morgan Stanley Conflict: Warsh worked at Morgan Stanley 1995-2002. During the 2008 crisis, as Fed Governor, he negotiated Morgan Stanley’s conversion to bank holding company status to access Fed emergency lending—a potential conflict that generates negative headlines.

Rick Rieder (28% Odds)

We believe Rieder’s odds are vastly over estimated—Rieder has never held a federal position, testified before Congress, or navigated Senate confirmation.

Rieder was head of Lehman’s Global Principal Strategies team (proprietary trading) until the September 2008 bankruptcy. Democrats will likely grill him on his involvement in this.

The Washington Examiner reported January 20 that Rieder has supported higher corporate tax rates and increased immigration—positions directly contrary to Trump’s agenda.

Kevin Hassett (5% Odds)

Hassett’s odds crated when Trump’s January 16 statement appeared to eliminate Hassett, though his “I want to keep you where you are.”

Five Scenarios Where Waller Wins

Scenario 1: Confirmability Crisis (20-25% probability) Warsh and Rieder hit serious confirmation turbulence. Bessent and Wiles argue Waller is the only candidate with proven bipartisan support.

Scenario 2: Labor Market Narrative (15-20%) Weak January/February jobs data shifts Trump’s framing from “Fed Chair who cuts rates” to “Fed Chair who protects jobs.” Waller’s soft landing prediction provides credibility.

Scenario 3: Powell Stay-On (25-30%) Powell announces he’ll remain as Governor through January 2028. Waller’s working relationship with Powell (both dissented in July 2025) provides operational compatibility other candidates lack.

Scenario 4: The Tillis Trade (15-20%) Tillis negotiates: he drops the blockade if Trump withdraws the Powell investigation and nominates a genuinely independent Chair. Waller’s “absolutely” defense of Fed independence provides Tillis cover.

Scenario 5: Bessent Advocacy (10-15%) Treasury Secretary privately pushes Waller as optimal for confirmability, competence, and dovish policy. Bessent’s market credibility gives him unique influence over Trump’s decision.

The Bear Case

Despite structural strengths, Waller faces headwinds:

• Trump historically prioritizes personal relationships and media presence over technocratic competence

• Waller’s December 17 statement that he would “absolutely” defend Fed independence may have been disqualifying

• Sustained 12% odds despite his December interview suggest insiders know he’s not Trump’s choice

• At 67 in 2026, Waller may be seen as too old for a decade-shaping appointment

The Contrarian Position:

We believe Waller at 12% represents significant value. A $1,000 bet returns $8,300 if he wins.

Please bear in mind— the above statement is governed via our Terms

🔒 Early Warning Indicators (Paid Subscribers Only) 🔒

The following intelligence collection priorities and early warning signals are available to paid subscribers. These indicators may provide 24-72 hours advance notice of the selection before public announcement.

Read more

A recently closed mention market on Kalshi just delivered a masterclass in why you should read the fine print. I’ve said this numerous times, on pretty much all of my Kalshi / Event-Contract articles.

READ.

THE.

RULES.

The setup was simple: bet on whether a speaker would say “Warner Bros” during a public address. The speaker mentioned “Warner Brothers.” The contract is now in limbo—and traders are learning an expensive lesson about the difference between common sense and contractual language.

The Literal vs. The Obvious

On its face, this feels absurd. Warner Bros is Warner Brothers. Same company. Same referent. If you asked anyone on the street whether the speaker “mentioned Warner Bros,” they’d say yes without hesitation.

Again, this is why it is absolutely vital that you READ THE RULES.

The rulebook is explicit about what counts:

• ✅ Plurals and possessives — “Immigrants” counts for “Immigrant”

• ✅ Hyphenated compounds — “pro-Palestine” counts for “Palestine”

• ✅ Homonyms — “ice water” counts for “ICE”

• ❌ Grammatical inflections — “Immigration” doesn’t count for “Immigrant”

• ❌ Homophones — “right” doesn’t count for “write”

Notice what’s missing? Any provision for abbreviations and their expansions.

The NO Case is Stronger

The rules use slashes to indicate alternatives: “Elon / Musk” and “Doge / Dogecoin” appear as examples. If Kalshi intended both “Warner Bros” and “Warner Brothers” to resolve the market, they had a clear mechanism to say so.

They didn’t. And I will play devil’s advocate that Kalshi should’ve included the / in the market and named it Warner Bros / Warner Brothers… but they didn’t.

With that in mind the YES side will argue intent, common understanding, and the “spirit” of the market. They’ll point to the transliteration rule allowing variations of “Zelensky” and claim this is analogous.

But here’s the problem: “Zelenski” and “Zelensky” are spelling variations of the same word. “Bros” and “Brothers” are different words entirely—one is an abbreviation, the other is the full form. The rules are strict everywhere else. Why would this be the exception?

The Precedent Problem

If “Warner Brothers” counts for “Warner Bros,” where does it stop?

• Does “United States” count for “US”?

• Does “okay” count for “OK”?

• Does “cryptocurrency” count for “crypto”?

Markets need predictable rules. The moment you start adjudicating based on “reasonable interpretation” rather than textual analysis, you’ve created uncertainty that makes these contracts harder to price.

What Happens Next

Expect a support ticket war. Traders on both sides will make their case, and Kalshi will have to rule.

My read: the NO side should prevail on a strict textual interpretation. But this is Kalshi—outcomes aren’t always predictable, and “common sense” arguments have swayed resolutions before.

In very rare cases, Kalshi reserves the right to “wash” the market and void every parties trade. This would result in all shares being “burned” and allocated capital returned to traders—granted they held to expiration.

Either way, this market is a reminder: in prediction markets, the edge isn’t just in forecasting events. It’s in forecasting how the rules will be applied to those events.

What’re your thoughts? Do you think YES or NO holders should be paid?

Leave a comment

This article is for informational and educational purposes only and does not constitute investment advice, trading advice, or a recommendation to buy, sell, or hold any position in any event contract or financial instrument. Event contracts traded on CFTC-regulated exchanges like Kalshi involve risk, including the potential loss of your entire investment. Past performance is not indicative of future results.

Readers should conduct their own research and consult with a qualified financial advisor before making any trading decisions. The views expressed are solely those of the author and do not represent the views of any exchange or regulatory body.

Event contract rules and settlement outcomes are determined solely by the relevant exchange. The author’s interpretation of contract rules is opinion only and should not be relied upon for trading decisions.

Position Disclaimer: The author of this article holds NO “Warner Bros” contract(s.)