Free cash flow: the number that's hardest to fake

August 5, 2026 · 7 min readfundamentalseducation

There's an old line on trading desks: earnings are an opinion; cash is a fact. Like most old lines it oversimplifies — we'll get to how — but it points at something real about how financial statements are built. Reported earnings are the output of hundreds of judgment calls: when revenue counts as earned, how fast machines depreciate, which costs get spread across future years. Free cash flow is closer to a bank statement. Money came in, money went out, here's the difference.

That's why free cash flow — FCF — is the first number worth learning to read for yourself, and the first one worth building a screen around. This post covers what it is, where it lives in the filings, why it diverges from earnings, one famous case where the divergence was the whole story, and how you'd turn it into a testable screen.

What it is, from the actual filings

Free cash flow isn't a line on any statement. It's a two-term calculation from the cash flow statement, which every US public company files quarterly (10-Q) and annually (10-K):

free cash flow = operating cash flow − capital expenditures

Operating cash flow (OCF) is the total of the statement's first section: cash the business actually collected and paid in the course of operating. It starts from net income, adds back non-cash charges like depreciation, and adjusts for working capital — receivables billed but not collected, inventory bought but not sold, bills incurred but not paid.

Capital expenditures (capex) sits in the investing section, usually titled "purchases of property, plant and equipment": cash spent on long-lived assets.

Subtract one from the other and you get the cash the business generated after paying to maintain and grow its asset base — the money genuinely available to pay down debt, buy back stock, pay dividends, or pile up. When investors talk about a company "gushing cash," this is the number they mean, whether they know it or not.

Both inputs come straight from the primary documents. You can open any company's 10-K on SEC EDGAR and compute FCF yourself with fifteen seconds of arithmetic — no vendor adjustments, no methodology page. That "check it yourself" property is the spirit of the whole exercise.

Why earnings and FCF diverge

If earnings measured cash, the two numbers would match. They don't, because earnings are built on accruals — accounting entries that shift income recognition in time. Revenue is booked when earned, not when collected; costs are booked when matched to revenue, not when paid. Accrual accounting exists for good reasons; it genuinely smooths timing noise. But every accrual is a judgment, and judgments bend under pressure.

Divergence between earnings and FCF has innocent and guilty explanations, and telling them apart is the skill:

Innocent: a genuinely growing company consumes cash — inventory ahead of sales, receivables from new customers, heavy capex for capacity. Earnings positive, FCF negative, story checks out if the growth is real and the gap narrows as it matures.
Less innocent: receivables growing much faster than revenue (booking sales customers haven't paid for — sometimes never will), inventory swelling against flat sales, or costs being capitalized — recorded as asset purchases instead of expenses, which flatters earnings today and quietly hollows out the future.

One quarter of divergence is noise. Years of rising earnings with stagnant or negative free cash flow is a pattern, and it has preceded a remarkable share of the market's great blowups.

The famous one: WorldCom

The clearest historical illustration is WorldCom, and it's worth telling from the public record because of which numbers the fraud could and couldn't reach.

In June 2002, WorldCom — then one of the largest telecoms in the world — announced that it had improperly transferred billions of dollars of "line costs" (fees paid to other carriers to complete calls) from operating expenses to capital accounts; the SEC's fraud complaint followed within days, and the restatements ultimately ran to roughly $11 billion. The company filed what was then the largest bankruptcy in US history the following month.

Mechanically, the trick was reclassification. Booking an operating expense as capex does two flattering things at once: the expense vanishes from the income statement, inflating reported profit — and the cash outflow moves from the operating section to the investing section, inflating reported operating cash flow too. Both headline numbers looked healthier. Earnings were "an opinion," and so, it turns out, was OCF.

But look at the arithmetic of free cash flow: OCF went up by the misclassified amount, and capex went up by the same amount. FCF = OCF − capex was left essentially untouched. The reclassification could dress up earnings and even operating cash flow; it could not manufacture free cash — every dollar smuggled into capex got subtracted right back out. A reader tracking the spread between WorldCom's reported profits and its free cash flow was watching the one gauge the scheme couldn't easily move.

That's the sense in which FCF is the number that's hardest to fake: the two components sit in different sections of the statement and subtract against each other, so the most tempting cosmetic — moving costs below the operating line — cancels out.

Hardest ≠ impossible

Honesty requires the caveats, because FCF can be bent too — mostly through timing:

Stretching payables (paying suppliers slower) boosts OCF for a while. It shows up as ballooning payables and it borrows from next year's number.
Lumpy or starved capex distorts single periods — a company can flatter this year's FCF by underinvesting, which is a real decision with a real cost that arrives later. Use trailing-twelve-month figures and look across several years.
Stock-based compensation is added back to OCF as a non-cash expense, but it pays employees with dilution — a genuine cost FCF doesn't charge. For heavy issuers, look at FCF per share over time, not just FCF.
Selling receivables (factoring) pulls future collections into today's OCF.

So the practical posture isn't "FCF is truth"; it's "FCF is the hardest number to fake for long, and its divergence from earnings is the most informative red flag in the statements." Persistent, unexplained divergence is the tell — in either direction.

Turning it into a screen

Reading one company's cash flow statement is diligence; the leverage comes from applying the same discipline to the entire market at once. In Quantery you'd express it from the raw point-in-time fundamentals — the shape below is illustrative (the bundled templates are the reference for exact field names), but this is the whole idea:

# "Real cash, fairly priced, earnings backed by it" — illustrative
params:
  yield_strong: 0.06         # FCF ≥ 6% of market cap
  yield_ok:     0.03
  conversion_floor: 0.8      # FCF should roughly back reported earnings

features:
  ocf_ttm:    ttm(operating_cash_flow)
  fcf_ttm:    ttm(free_cash_flow)
  ni_ttm:     ttm(net_income)
  fcf_yield:  if(market_cap > 0, fcf_ttm / market_cap, null)
  conversion: if(ni_ttm > 0, fcf_ttm / ni_ttm, null)   # the WorldCom gauge

criteria:
  cash_generation:
    - { when: "fcf_yield >= $yield_strong", score: 2 }
    - { when: "fcf_yield >= $yield_ok",     score: 1 }
    - { else: 0 }
  earnings_quality:
    - { when: "conversion >= $conversion_floor", score: 2 }
    - { when: "conversion >= 0.5",               score: 1 }
    - { else: 0 }

gate:
  - fcf_ttm > 0              # no credit for projected cash

The conversion feature is the point of the whole post compressed into one line: trailing free cash flow divided by trailing net income. Near or above 1, the earnings are cash-backed; persistently far below, you've found the divergence — and your screen found it across every filer in the universe, on point-in-time data, while you were doing something else. Because every criterion in the survivors table shows its number and source, a high scorer isn't an assertion — it's an arithmetic claim you can check against the filings it came from.

Screenshot pending

Survivors table for the FCF screen, with the cash_generation and earnings_quality criterion scores visible

The screen's survivors — each row's yield and conversion numbers one click from the filings behind them.

None of this makes high-FCF companies automatic buys — a business can gush cash straight into decline, and the market often prices cash gushers fairly. What it gives you is a market-wide reading of the one number that's hardest to fake, computed your way, with the receipts attached. Whether that becomes a thesis worth money is what backtesting is for.

Research tooling, not investment advice. Nothing here is a recommendation to buy, sell, or hold any security. Screens, scores, and backtests are informational only; backtested results are hypothetical, exclude costs such as commissions and slippage, and do not guarantee future results. Verify against primary filings and make your own decisions.

Want to try this on your own rules? Quantery is free for 14 days — the full app, no card required.

← All articles