Filing Flows › Companies › Graham Holdings › Analysis traps
Graham Holdings (GHC): analysis traps
Where its figures could mislead a reader, and what to check before taking them at face value · Data: Q3 2026, to Sep 30, 2026, filed Oct 29, 2026; rules v1.1
Graham Holdings (GHC): 0 rules of 10 triggered on the filings to Sep 30, 2026. 1 to review; 2 with insufficient data; 2 not applicable. Each is a question to verify, with the normal explanation that would produce the same figures.
Check these first
A precise DCF value = the true value To review · confidence high
The DCF's point value is 625.92 a share (price 1,212.02); 76% of it is the terminal value, a number about years no one has seen. Moving operating margin by ±3 pts alone takes it from 136.92 to 1,114.92; across every method the research ran the range is -1,474 to 6,357. The value is a range; the decimals are not information.
Innocent reading: A stable, mature business with a narrow range of outcomes, where the point value is a fair summary.
Every rule
A01 FCF growth = the business improved Not triggered confidence high
Free cash flow did not rise ($324M → $275M): the trap does not arise.
The figures and how they are read
| Fiscal year before | FY2024 |
| Fiscal year | FY2025 |
| Operating cash flow, year before | $407M |
| Operating cash flow | $347M |
| Capex, year before | $83M |
| Capex | $72M |
| Free cash flow, year before | $324M |
| Free cash flow | $275M |
| Change in FCF | −$49M |
| Change in capex | −$11M |
| Working capital: change in its cash contribution | −$92M |
FCF = operating cash flow − cash capex; ΔFCF = ΔOCF − ΔCapEx; the working-capital part is the change in the cash-bridge items (receivables, inventory, payables, accruals), with the sign of the cash flow statement
Source: FY2025 (cash flow statement, fiscal year) · FY2024 (cash flow statement, fiscal year before)
A07 Stock compensation added back = no economic cost Not triggered confidence high
Stock compensation is 0.1% of revenue: small enough that adding it back changes little.
The figures and how they are read
| Fiscal year | FY2025 |
| Stock compensation | $6M |
| Revenue | $4.9B |
| SBC / revenue | +0.1% |
| Buybacks | $3M |
| Buybacks / SBC | 0.6 |
| Diluted shares, year before | 4.4M shares |
| Diluted shares | 4.4M shares |
| Diluted shares: change | -0.7% |
SBC / revenue from the cash flow statement's add-back; the diluted weighted share count year over year; buybacks from the financing section
Source: FY2025 (stock compensation, buybacks and the diluted share count, fiscal year)
B02 Receivables growing = future cash is certain Not triggered confidence medium
Receivables grew 0% against revenue 4% in Q3 2026; DSO 32 against a median of 36: collections are keeping pace with sales.
The figures and how they are read
| Quarter | Q3 2026 |
| Revenue, year over year | +3.7% |
| Receivables, year over year | -0.5% |
| Quarter before | Q2 2026 |
| Revenue y/y, quarter before | +7.1% |
| Receivables y/y, quarter before | +0.0% |
| Days sales outstanding | 31.6 |
| DSO, median of eight quarters | 35.6 |
| Receivables | $492M |
| Revenue | $1.3B |
receivables and revenue against the same quarter a year earlier; DSO = receivables / quarterly revenue × days, against the median of the eight quarters before
Source: Q3 2026 (receivables and revenue) · Q3 2025 (the same quarter a year earlier)
B08 Adjusted profit = economic profit Insufficient data
no reconciliation table read from the releases yet and fewer than three quarters with both an adjusted and a GAAP EPS
C02 Low net debt = no refinancing risk Insufficient data
no balance-sheet cash or debt on file
D03 EPS growth = profit growth Not triggered confidence high
EPS fell 37% and net income fell 39% in Q3 2026; the share count fell 3.1%: EPS growth follows profit growth.
The figures and how they are read
| Quarter | Q3 2026 |
| Quarter before | Q3 2025 |
| EPS, year before | $28.08 |
| EPS | $17.70 |
| Net income, year before | $123M |
| Net income | $75M |
| Diluted shares, year before | 4.4M shares |
| Diluted shares | 4.2M shares |
| EPS growth | -37.0% |
| Net income growth | -38.9% |
| Diluted shares: change | -3.1% |
| eps from profit | -$10.92 |
| eps from share count | $0.55 |
EPS and net income against the same quarter a year earlier; the EPS bridge attributes the change to profit first, then to the share count (the order is fixed and disclosed)
Source: Q3 2026 (net income, diluted shares, EPS) · Q3 2025 (the same quarter a year earlier)
E02 Low P/E at a cyclical peak = undervalued Not triggered confidence medium
Net income of $542M is not at a cyclical high against the last eight years (median $300M, best $725M): the P/E of 9.5× is on a normal year.
The figures and how they are read
| market value | $5.1B |
| net income ttm | $542M |
| pe ttm | 9.50 |
| net income median fy | $300M |
| pe mid cycle | 17.10 |
| net income peak fy | $725M |
| years | FY2018 / 271.0; FY2019 / 328.0; FY2020 / 300.0; FY2021 / 352.0; FY2022 / 67.0; FY2023 / 205.0; FY2024 / 725.0; FY2025 / 292.0 |
| down years in window | 3 |
P/E on the trailing twelve months' net income against the P/E on the median net income of the last eight fiscal years (a mid-cycle stand-in); 'at a peak' = this year's profit at or above 90% of the best year and at least 1.3× the median
Source: Q3 2026 (net income, trailing twelve months)
E07 A precise DCF value = the true value To review confidence high
The DCF's point value is 625.92 a share (price 1,212.02); 76% of it is the terminal value, a number about years no one has seen. Moving operating margin by ±3 pts alone takes it from 136.92 to 1,114.92; across every method the research ran the range is -1,474 to 6,357. The value is a range; the decimals are not information.
The figures and how they are read
| dcf value | 625.92 |
| price | 1,212.02 |
| terminal value share | +75.5% |
| swing by input | Operating margin / ±3 pts / 136.92 / 1114.92; WACC / ±1 pt / 499.13 / 813.67; Revenue growth (each year) / ±5 pts / 490.65 / 788.02; Capex (share of revenue) / ±1 pt / 571.45 / 680.39 |
| range all methods | -1474.23; 6357.49 |
the DCF's point value against the share of it that is the terminal value and the value at each input moved by one step (growth ±5 pts, WACC ±1 pt, margin ±3 pts …); the range across every method the research ran
Source: 2026-11-08 (valuation research)
What would explain it innocently
- A stable, mature business with a narrow range of outcomes, where the point value is a fair summary
- The inputs that matter most are themselves well known (a regulated return, a long contract)
What to look for in the filing
- Which input you believe differs from the research's and why
- What the price implies for growth (the reverse DCF on the fair-value page)
What would show the trap does not apply
- The DCF's value is robust to the terminal growth and discount rate (the range's low end is still above the price)
- The company's cash flows over the next three years track the DCF's early years
Look again: next filing.
J06 A bank's low P/E and high profit = safe Not applicable
not a bank
J08 A REIT's low net income = the dividend is unsafe Not applicable
not a REIT
What the statuses mean
Triggered: the filings show the pattern; read the normal explanations and the missing evidence before concluding anything. To review: one quarter or a weak signal; a reader's judgement, not a pattern. Not triggered: the pattern is absent on these figures. Insufficient data: the filings read so far do not hold what the rule needs; this is not a clean bill. Not applicable: the rule is not meant for this kind of company (a bank's working capital, a loss-making quarter's EPS growth). Out of date: the latest filing is more than fifteen months old.
Severity is what the pattern could mean for cash, value or solvency; confidence is how reliable the inputs and the reading are. The two are kept apart.
Record a view
A reader's judgement, grounds, what would change it and the next review; kept in the browser (needs JavaScript).
Every rule and the companies it triggers on → Graham Holdings: cash flow → Earnings quality score →
Each rule reads the company's own filings by a fixed formula and says what would explain the same figures innocently. A triggered rule is a question to verify, not a finding of manipulation, and not a view on the stock. Missing data is shown as missing, never as clear. Not investment advice.