Playbook › Playbook
A seasonal peak quarter's FCF read as a TTM figure inverts the cash-flow trend
Claim
A "TTM" figure must be the sum of four quarters. If it was read off one column, it is not TTM, and against a seasonal business the error always points the same way — down.
Intuit's fiscal year ends July 31 and its cash flow is violently seasonal, because tax season lands in fiscal Q3 (February–April). The quarterly FCF series at 2026-08-04 was:
| Quarter | Period end | FCF |
|---|---|---|
| Q4 FY25 | 2025-07-31 | $356M |
| Q1 FY26 | 2025-10-31 | $599M |
| Q2 FY26 | 2026-01-31 | $1,524M |
| Q3 FY26 | 2026-04-30 | $5,236M |
| True TTM | $7,715M |
The $5,236M cell is one quarter — the tax-season quarter, which alone carries about two-thirds
of the year's cash generation. Recorded as "TTM FCF ~$5.23B" and compared against FY2025's
$6.08B, it produced a 14% decline where the truth was a 41% increase ($6.08B → $8.617B
for FY2026, from the company's own release: OCF $8.838B less capex $221M).
Why it survived review
The baseline report was careful in every other respect — it voided DYT on a known pitfall, refused to estimate SBC rather than inventing it, and explicitly flagged this very figure as "not yet resolvable and must not be waved away," running the valuation on both the $6.08B and $5.23B cases. The discipline was right and the input was wrong, which is the dangerous combination: a flagged uncertainty reads as handled, so nobody re-derives the number underneath it.
The magnitude also looked plausible. A 14% FCF dip is unremarkable; had the misread produced a absurd figure it would have been caught. Seasonal misreads are hard to spot precisely because one quarter of a strongly seasonal year is a believable full-year number.
The tell
A TTM figure that exactly matches a single cell in the quarterly table is not a coincidence. Here "$5.23B TTM" versus the Q3 column's $5.236B agreed to three significant figures. Any TTM that reproduces one quarterly cell to that precision was copied, not summed.
What it contaminates
The error does not stay in one row. It flowed into:
- The cash-flow trend — inverted, decline for growth.
- The dividend FCF payout ratio — computed as
$1.31B ÷ $5.23B ≈ 25%when the real denominator gave ~17%. The conclusion ("survives a 75% cut to FCF") was directionally right but understated; the true cushion was a ~83% cut. - Any FCF-multiple valuation built on the same denominator.
The rule
- Never take TTM from a column. Sum four quarters, and show the addition.
- Check the sum against a single quarter's value — if they match, you copied.
- For seasonal businesses, name the seasonal quarter before computing anything: Intuit's Q3 (tax season), retailers' Q4, insurers' catastrophe quarters. The peak quarter will impersonate a year.
- When a full-year figure lands, reconcile it back to the quarterly sum. FY2026 OCF $8.838B less capex $221M reconciles to the four quarterly cells; the $5.23B never did.
Related: [[pitfall-fin-py-snapshot-mixes-annual-and-mrq-dates]] is the neighbouring failure — that one mixes as-of dates between blocks, this one mixes periods within one block. Both produce a number that is individually real and contextually false.