IPOs are minefields. You're buying a story before there's a track record, and the people selling you that story have every incentive to make it sound better than it is.
That's not cynicism — it's just the mechanics. The founders want the highest price. The VCs need to show returns to their LPs. The investment banks get paid on deal size. The retail investor shows up last, after everyone with better information has already priced in their edge.
Why Build a Framework?
When you analyze an established public company, you have years of earnings calls, quarterly reports, and market-tested valuations. The information asymmetry between you and insiders is smaller (though never zero).
IPOs flip this. The S-1 prospectus is a 200-page legal document where companies are required to disclose their risks — and strategically incentivized to bury them. The valuation was set in a negotiation between bankers and institutions, not by open-market price discovery.
The Six Stages
The Gauntlet forces systematic analysis across six dimensions before you can reach a classification. Each stage asks specific questions and produces specific outputs.
Stage 1: S-1 Forensics
Extract the buried signals from the prospectus. Non-boilerplate risk factors, related-party transactions, use of proceeds. 'General corporate purposes' for 80% of the raise? Yellow flag.
Stage 2: Unit Economics Reality Check
Normalize for stock-based compensation. A SaaS company reporting 'Adjusted EBITDA' usually means 'EBITDA if we didn't pay employees in stock.' Those options dilute you eventually.
Stage 3: Governance & Alignment
Who is the company being run for? Dual-class voting, related-party deals, compensation packages that pay out regardless of performance. Not automatic disqualifiers, but signals.
Stage 4: Supply Mechanics
Lock-up expirations flood the market with shares from investors whose cost basis is near zero. They're not selling because the business is bad — they're selling because they finally can.
Stage 5: Valuation Sanity
'Down 50% from IPO' doesn't mean cheap. The IPO price was a negotiation, not a market-clearing price. Reverse-engineer what growth assumptions are baked into current prices.
The Classification System
After working through the stages, you end up with one of six classifications. This isn't about predicting stock prices — it's about knowing what you're dealing with.
Early Compounder
High-quality business, reasonable valuation, aligned insiders. These are rare.
Highest conviction level
Strong Business, Unfavorable Setup
Quality company, but valuation is stretched or a supply event (like lock-up expiration) is imminent.
Define specific conditions that would change the setup
Promising, but Unproven
Interesting business model, but not enough post-IPO data to validate the thesis yet.
Requires 1-2 quarters of public results
Story Over Substance
Valuation requires heroic assumptions. The narrative is doing more work than the numbers.
High risk of re-rating when reality doesn't match expectations
Exit Liquidity Trap
IPO primarily benefits early investors. Weak economics, extractive governance, or both.
Structural problems unlikely to resolve
Insufficient Data
Critical information missing. Can't form a high-conviction view either way.
Analysis blocked until disclosure improves
What This Framework Won't Do
We want to be clear about limitations:
- It won't predict prices. Markets are hard. Anyone promising systematic outperformance from a checklist is selling something.
- It won't work for SPACs. Different mechanics entirely — warrants, redemptions, sponsor promotes. Requires separate framework.
- It won't work for pre-revenue biotech. You're underwriting clinical science, not business models. Outside our competence.
- It won't replace due diligence. The framework is the work, not a substitute for it.
How We Built It
We ran proposals through four different LLMs (Claude, ChatGPT, Gemini, Perplexity), then had each critique the others' work. The disagreements were informative:
Opus pushed for more explicit checklists and decision trees — "too easy to skip a step without them."
Gemini argued for sector-specific modules — "you can't analyze SaaS and industrials with the same metrics."
ChatGPT wanted quantitative thresholds for overhang risk — "if you can't put a number on it, you'll rationalize it away."
Perplexity emphasized time-phased analysis — "a 3-month-old IPO needs different emphasis than a 15-month-old one."
The final framework incorporates all of these. The multi-model process produced something more rigorous than any single pass would have.
The Prospectus Probe is now available
We run every IPO candidate through the full framework before forming a view. Analyses appear in our equity research section as they're completed.
Explore IPO Analyses →Further Reading
Prospectus Probe — Full Reference
Complete lens documentation: signals produced, patterns detected, and how it fits with other lenses.
Fugazi Filter
Our lens for detecting accounting manipulation and financial engineering.
Evidence Ladder
How we grade the quality of evidence supporting each finding (E0-E3).
LLM-isms
Patterns that suggest shallow analysis, and how we try to avoid them.