Reading SEC filings

Which filing answers which question

Using EDGAR like a professional

• Company search at sec.gov/edgar: search by name or ticker, then filter by form type. Bookmark the company's filing index; it is the primary source everything else quotes.
• Full-text search (efts.sec.gov/LATEST/search-index or the EDGAR full-text UI) searches INSIDE filings since 2001: every mention of a competitor, a customer name, or a phrase like "going concern" across all companies. One of the most underused research tools that exists, at any price.
• Compare year over year. Open this year's and last year's 10-K side by side and read what CHANGED: new risk factors, reworded accounting policies, a segment quietly redefined. Companies edit these documents carefully; the edits are the signal.
• The exhibits index at the back lists the actual contracts: credit agreements, executive employment terms, major customer agreements (sometimes redacted). When a debt covenant matters, the real document is Exhibit 10-something, not the summary.

The reporting calendar

The calendar itself carries signal: a company filing an NT 10-K ("notification of late filing") is announcing that something, often accounting review, prevented it from closing its books, and the market treats late filings as the smoke they usually are.

Inside a 10-K: the sections that matter

A 10-K can run three hundred pages and most is boilerplate. Four sections carry nearly all the value:

• Item 1, Business. The company describing itself: segments, products, customers, competition, regulation. The first place to learn how it actually makes money, and where customer concentration ("one customer accounted for 22% of revenue") is disclosed.
• Item 1A, Risk Factors. Mostly legal armor, but the SPECIFIC risks (a named customer, a patent expiring on a date, one supplier for a key part, a covenant level) are disclosures found nowhere else. Skim the generic; hunt the specific; and compare against last year for NEW entries, which are new for a reason.
• Item 7, MD&A. Management's Discussion and Analysis: management explaining, with numbers, why results moved. The single highest-value section, because it decomposes changes into price, volume, currency, and acquisitions, which no summary service does faithfully. The liquidity subsection (cash needs, maturities, covenants) is the credit picture in prose.
• Item 8, Statements and Notes. The audited statements plus footnotes: debt maturity schedules, lease obligations, segment detail, pension assumptions, litigation, revenue-recognition policy, and the auditor's Critical Audit Matters: the auditor naming, in public, the estimates it found hardest to verify. Read CAMs first; they are a map of where judgment lives.

A worked MD&A read

"Revenue increased 9.4% to $2,340M. On a constant-currency basis
revenue grew 11.2%, comprising 6.8% volume growth, 3.1% from pricing,
and 1.3% from the Delta acquisition, partially offset by a 1.8%
currency headwind."

the analyst's extraction:
  organic growth        = 6.8 + 3.1            = 9.9%   <- the real number
  acquired growth       = 1.3%                          <- bought, not built
  FX                    = -1.8%                         <- noise, mean-reverting
  price vs volume mix   = pricing power is 1/3 of organic: check if price
                          held while volumes grew (healthy) or replaced them

That decomposition is why MD&A exists. Headline growth of 9.4% could hide a shrinking business padded by deals; here it hides a BETTER business than the headline. No screener field contains this; it is prose, and reading it is the edge.

8-Ks: the item codes worth knowing

The proxy: pay and power

The DEF 14A answers the question every other filing avoids: what is management paid to do? The compensation discussion lays out the bonus metrics and their weights, and people deliver what they are paid for: bonuses on revenue produce acquisitions; on EPS, buybacks at any price; on relative TSR or ROIC over multi-year periods, discipline. The proxy also shows ownership (founders with real skin versus hired hands with granted options), related-party transactions, and the board's actual independence. Ten minutes in the proxy predicts capital allocation better than an hour of interviews.

Filings versus earnings calls

The quarterly triangle is release, call, then 10-Q, and each has a different honesty gradient. The release is marketing with numbers; the call's prepared remarks are rehearsed, but the Q&A is the live document: which questions management dodges, which analyst pushes, whether guidance language shifts from "confident" to "committed to improving." The 10-Q arriving weeks later contains what neither mentioned. The professional habit: note the claims made on the call, then verify them against the Q when it lands. The gaps are a running honesty score for the management team.

A working reading order for a new company

• Item 1 of the latest 10-K: learn the business from the source.
• MD&A for the last two years: what actually drives results, decomposed.
• The statements and notes: debt schedule, segment note, revenue recognition, CAMs.
• The proxy: how management is paid, and who owns what.
• The last four 8-Ks and the newest 10-Q: what changed since the annual picture.
• Then, only then, the earnings call transcripts, with claims to check against all of the above.

What to read with gloves on

• Non-GAAP measures. "Adjusted EBITDA" excludes whatever the company chooses. The filing must reconcile it to the audited number; always read the reconciliation, because the SIZE and GROWTH of the gap is the story. An adjustment that recurs every quarter is a cost, renamed.
• Quiet accounting-policy changes. A changed depreciation life or revenue-recognition method can manufacture a year of growth; the policy note discloses it in one dry sentence.
• Segment redefinitions. Re-slicing segments resets every trend line an analyst was tracking. Sometimes legitimate; always convenient.
• The word "substantially." Legal drafting hides precision. When a filing goes vague where a number was possible, the vagueness was chosen.