The 200-Page Confession

Every public company tells you exactly what could destroy it. Filed with the SEC, in plain English, on a free government website. The annual 10-K filing lists every material risk the company’s lawyers think could hurt shareholders. Pending lawsuits. Supplier dependencies. Revenue concentrations. Regulatory threats. Debt covenants approaching breach. All there.

The catch: it’s buried in 200-page documents that most investors never open. Even sell-side analysts often skim the financial tables and skip the risk factors. Retail investors read the headline numbers on Yahoo Finance. The gap between what’s disclosed and what’s priced is where grey swan opportunities live.

This kind of pattern-matching across filings used to take a team of analysts and weeks of reading. It’s faster now with searchable databases and AI tools that can compare documents. But knowing what to look for still matters more than the tools you use to look.

EDGAR: The Free Database Nobody Uses Enough

EDGAR is the SEC’s filing database. Every public company filing since the mid-1990s lives there. Full text. Searchable. Free.

Most investors know EDGAR exists. Almost none use it effectively. They type a company name, find the most recent 10-K, maybe skim the financial statements, and leave.

The power is in searching across filings, not just within one company. EDGAR has a full-text search API, and instead of reading one company’s disclosures, you can search for a pattern across every company at once:

  • Search “material weakness” across all 10-K filings in the past year. Every company that admitted an internal control problem shows up in one list.
  • Search “going concern” in the most recent quarter. Every auditor who questioned whether a company can survive the next 12 months.
  • Search “covenant breach” or “covenant waiver” in recent 8-Ks. Every company renegotiating with its lenders.
  • Search “restatement” in 8-K/A amendments. Every company correcting previous financial statements.

I also keep an eye on filing frequency per company. A company that normally files one 8-K per month and suddenly files four in a week is telling you something. That pattern is invisible if you’re only reading filings after someone else flags them.

Where to Look in a 10-K (Quickly)

The 10-K is the most comprehensive disclosure any public company makes. It’s also long. Here’s where to focus when you’re scanning rather than doing a full read:

  • Item 1A (Risk Factors): every material risk management wants shareholders to know about. Boilerplate in most years, but the changes between years are the signal. More on this below.
  • Item 3 (Legal Proceedings): pending litigation, regulatory actions, environmental claims. Often a single paragraph that understates material risk. Compare this to actual court docket data (Guide 4) to see if the company is underselling.
  • Item 7 (MD&A): management’s narrative about the business. Read for what they’re not saying as much as what they are. If a known risk exists and MD&A doesn’t address it, that silence is informative.
  • Item 8, Footnotes: commitments, contingencies, off-balance-sheet arrangements, customer and supplier concentration. The numbers in the tables are already priced. The footnotes often aren’t. Specifically: Note on “Commitments and Contingencies” and Note on “Segment Information” or “Geographic Information” for revenue and cost concentration.

If I’m moving fast, I Ctrl+F for “commitment,” “minimum,” “concentration,” “related party,” “impairment,” and “restatement” in the footnotes. Those six words surface the disclosures that routinely get missed: purchase obligations, customer concentration thresholds, related-party transactions, asset writedowns, and corrections to prior financials.

This isn’t a complete map. But it’s enough to spend 30 minutes on a 10-K and walk away knowing whether the company has disclosed anything that contradicts the market’s pricing.

The Other Filing Types That Matter

10-Q: the quarterly update

Quarterly filings. Less comprehensive than 10-Ks but valuable for tracking changes. New risk factors appearing mid-year signal emerging problems that weren’t in the annual report. I compare 10-Q language quarter over quarter for companies I’m actively researching.

8-K: the real-time feed

8-K filings report material events as they happen. Companies must file 8-Ks for specific trigger events:

  • Item 1.01: entry into a material definitive agreement (new debt, amended credit facilities)
  • Item 2.04: triggering events that accelerate or increase a direct financial obligation (covenant breaches)
  • Item 4.02: non-reliance on previously issued financial statements (restatement incoming)
  • Item 5.02: departure of directors or certain officers (executive exits)
  • Item 8.01: other events the company deems material (catch-all)

The trouble signals: multiple 8-K amendments (the original disclosure was wrong or incomplete), rapid-fire 8-Ks (managing multiple events simultaneously), and late-Friday 8-Ks (hoping for minimal weekend coverage).

NT 10-K / NT 10-Q: the late filing notice

“NT” stands for “notification of late filing.” A company files an NT when it can’t meet the deadline for its annual or quarterly report.

I treat NT filings as a yellow flag worth investigating immediately. The reasons range from benign (merger complexity, accounting system transitions) to serious (restatement work, auditor disagreements, material weakness remediation). But in my experience, the base rate of bad outcomes is high enough that I don’t ignore them. A search for “NT 10-K” filings in the past year produces a surprisingly useful list of companies worth watching.

SC 13D: the activist arrival

When an investor acquires more than 5% of a company’s shares with intent to influence management, they file an SC 13D. Activist investors push for board seats, strategic reviews, asset sales, management changes. The stock moves on the filing, but the real action plays out over months.

DEFM14A: the merger proxy

Filed when a company is being acquired. Contains the board’s recommendation, fairness opinions, dissenting views, and the price negotiation history. Proxy fights over contested mergers can create forced selling and pricing dislocations.

Reading Risk Factors Like a Diff

Every 10-K has a risk factors section. Most of it is boilerplate that hasn’t changed in five years. “Our business may be adversely affected by general economic conditions.” Lawyers copy-paste these paragraphs annually.

The technique that transforms risk factors from noise to signal: diff this year’s against last year’s.

New language means new risk. Management’s lawyers don’t add risk factor paragraphs for fun. They add them because something changed and legal counsel determined it was material enough to disclose.

A concrete example from our tariffs research. A consumer goods company’s 10-K risk factors had included a generic paragraph about “international trade policies” for three straight years, essentially unchanged. Then in the most recent filing, three things changed:

  • The word “tariff” appeared 14 times (up from 2)
  • A new sentence named specific countries: “substantially all of our products are manufactured in Vietnam and Cambodia”
  • The phrase “may adversely affect” was replaced with “has adversely affected and is expected to continue to adversely affect”

That shift from “may” to “has” is the most important single word change in the document. The lawyers moved from hypothetical risk to acknowledged reality. It told you the tariff impact was already hitting the business, even though the earnings call that quarter was still talking about “managing through.”

The CYA caveat

Not every risk factor change signals genuine new risk. Sometimes legal teams add language as a precaution after seeing another company get sued for inadequate disclosure. A wave of cybersecurity lawsuits, for example, prompted hundreds of companies to add or expand their cybersecurity risk factors even if nothing had changed at their company.

How to tell the difference: timing and specificity. If a new risk factor appears right after a competitor was sued for that exact issue, it’s probably defensive. If it names specific countries, quantifies exposure, or uses language that shifts from “may” to “has,” it’s more likely reflecting an actual change in the business.

AI is good at this comparison work. Feed two versions of a risk factors section into a conversation, ask what changed, and you get a precise diff in seconds. What matters is knowing what the diff means, not generating the diff itself.

8-K Pattern Analysis

Individual 8-K filings matter. But 8-K patterns matter more.

Think of 8-Ks as a heartbeat monitor. Normal companies file them at regular intervals around predictable events. A sudden spike in filing frequency is the equivalent of an elevated heart rate.

Cluster detection

A company filing 4+ 8-Ks in a two-week window outside of earnings season is managing something. Each filing represents a separate material event. Multiple events in quick succession means things are changing fast. Compare to the company’s normal cadence to calibrate what counts as unusual.

Amendment chains

8-K/A (amendment) filings mean the original disclosure was incomplete or evolved. One amendment is normal. Two amendments to the same event are concerning. Three or more means the company is struggling to define the scope of a problem.

The signal I missed

Early in this work, I dismissed an 8-K/A chain on a small-cap as bureaucratic cleanup. Three amendments to a single 8-K over six weeks, each expanding the scope of a “previously disclosed” credit facility issue. I assumed it was clerical. It wasn’t. The company was progressively admitting that the covenant breach was worse than initially reported. By the third amendment, the credit facility had been restructured at punitive terms. The stock dropped 40% over the following quarter as the margin compression hit earnings.

In hindsight, the pattern was obvious: amendment chains aren’t clerical. They’re evolving disclosures of problems that are getting worse. Now I treat any 8-K with two or more amendments as a checklist item.

Earnings call + 8-K divergence

Sometimes what the CEO says on the earnings call contradicts what the 8-K discloses. The call is optimistic. The 8-K filed the same day discloses a covenant waiver, an executive departure, or a credit facility amendment. The call is marketing. The filing is legal disclosure. When they diverge, trust the filing.

Practical Workflow: Filing to Thesis

Here’s how this actually works in practice.

Step 1: Pick a macro theme. Tariff changes, interest rate shifts, regulatory actions, weather events, technology disruption. The context determines which companies face pressure.

Step 2: Search EDGAR broadly. Find companies mentioning the specific risk in recent filings. “Tariff” in 10-K risk factors. “Rate covenant” in 8-Ks. “Supply chain disruption” in 10-Qs. The breadth-first pass matters: I scan across hundreds of tickers on a given theme, then go deep on the ones that surface signals. Trying to do it one company at a time is too slow and too narrow.

Step 3: Diff risk factors for the top hits. Pull the current and prior-year risk factor sections. Compare them. New language about the macro risk you identified in Step 1 means the company knows it’s exposed.

Step 4: Check 8-K filing frequency. Is the company filing more frequently than normal? Any amendments? Any late-Friday dumps?

Step 5: Cross-reference. Check insider trades for selling clusters (Guide 3). Check court dockets for related litigation (Guide 4). Insiders selling while the company discloses new risks in its 10-K is a compounding signal.

Step 6: Check the options chain (Guide 1). The thesis is formed. Now find out if the trade exists: liquid enough, cheap enough, expiration covering the catalyst window.

The power is in combining these steps. No single filing tells the whole story.

Limitations

SEC filing analysis has real limitations:

  • Disclosure lag. 10-Ks are filed weeks or months after year-end. 10-Qs lag by 40-45 days. By the time you read the disclosure, the situation may have already changed. 8-Ks are faster (days), but they only cover specific trigger events.
  • Boilerplate overwhelm. Most risk factor language doesn’t change and doesn’t matter. The technique depends on finding the changes, but if you’re not comparing versions, you’re reading noise. A single year’s filing, read in isolation, is much less useful than two years compared.
  • Legal conservatism masquerading as signal. As noted above, not every new risk factor reflects a real change. Some are added defensively. Experience helps calibrate this, but I’ve chased signals that turned out to be CYA language more than once.
  • Small companies don’t file much. Micro-caps and OTC companies have fewer filing requirements and less detailed disclosures. The technique works best on companies that file comprehensive 10-Ks with detailed risk factors, which generally means mid-cap and above.
  • The market reads the same filings. On large-cap, well-covered companies, sell-side analysts read the 10-K. Your edge isn’t in reading it; it’s in reading it differently (diffing, cross-referencing with court data and insider trades). On small- and mid-caps with less analyst coverage, the filings genuinely get overlooked.

If You Only Do Three Things

  1. Diff the risk factors. Pull this year’s 10-K risk factors and last year’s. Feed them to an AI or compare them manually. The new language is the signal.
  2. Watch 8-K filing frequency. A company filing 4+ 8-Ks in two weeks, or filing 8-K amendments to the same event, is managing something that’s evolving.
  3. Read the footnotes, not just the tables. Commitments, contingencies, customer concentration, and geographic cost breakdowns live in the footnotes. The headline numbers are priced. The footnotes often aren’t.