A.7.4 Quality & F-Scores

Basic Prerequisites: 03 · Value & Profitability and Core Financial Ratios · Advanced Extensions (F-score).

---

1. Intuition & Practical Objective

Value says cheap, profitability says good - but neither directly asks the question a distressed-value investor most needs answered: is this cheap firm financially strong, or is its cheapness a symptom of distress? This page answers it with the composite instruments of quality - most famously the Piotroski F-score, and behind it Sloan’s accruals factor.

Joseph Piotroski’s 2000 insight was precise and powerful. A broad portfolio of high book-to-market (value) firms earns a premium on average, but it is a messy average: it contains both genuine bargains and financially distressed companies that keep falling. Piotroski’s F-score - a 0–9 sum of nine binary accounting signals (profitability, leverage/liquidity, operating efficiency) - separates the winners from the losers within the cheap universe. His headline numbers are striking and often quoted: applying the screen to high-B/M firms raised the mean return by at least 7.5% annually, and a strategy that buys expected winners and shorts expected losers earned about 23% per year over 1976–1996.

The deeper quality factor behind the F-score is accruals. Because earnings are accrued, management has discretion over when revenue and expense hit. Sloan (1996) showed high-accrual (low cash-flow) earnings are less persistent and earn lower subsequent returns - so earnings quality is itself a priced characteristic (Chan, Jegadeesh & Lakonishok 2006 confirm it adds return-predictive power beyond value). The F-score bakes this in directly: one of its nine signals is “operating cash flow exceeds ROA” - i.e. earnings backed by cash, not by accrual padding.

---

2. Mathematical Ground Truth & Derivations

The accrual identity - where quality lives. Net income and operating cash flow differ by accruals:

$$\text{NI}=\text{CFO}+\text{Accruals}.$$

High accruals mean earnings outrun cash - income that is recognized but not collected. Sloan (1996) shows such earnings revert; the accrual component is less persistent than the cash component. The accruals factor is therefore long low-accrual, short high-accrual firms.

The Piotroski F-score. Nine binary signals, each $=1$ if “good”, summed to a 0–9 score (definitions verified against the corpus paper):

Area	Signal	”Good” condition
Profitability	F_ROA	$\text{ROA}=\text{NI}/{\text{TA}}_{\text{begin}}>0$
	F_CFO	$\text{CFO}>0$
	F_AROA	$\Delta \text{ROA}>0$ (current vs. prior year)
	F_ACCRUAL	$\text{CFO}>\text{ROA}$ (earnings backed by cash)
Leverage / liquidity / funding	F_ALEVER	$\Delta (\text{LTD}/\text{avg TA})<0$ (debt fell)
	F_ALIQUID	$\Delta$ current ratio $>0$
	F_EQOFFER	no common-equity issuance in prior year
Operating efficiency	F_AMARGIN	$\Delta$ gross margin $>0$
	F_ATURN	$\Delta$ asset turnover $>0$

$$F=\sum _{i=1}^{9}1[{\text{signal}}_i\text{good}],F\in \{0,\dots ,9\}.$$

Piotroski applies it inside the high-B/M universe (the conditioning is the point): high-$F$ (7–9) firms are the financially strong cheap firms; low-$F$ (0–4) firms are the distressed ones to avoid or short. This is value + quality as one mechanical screen.

---

3. Computational Implementation - the full F-score screen, stdlib only

Runs on the standard library. It computes all nine Piotroski signals for a six-company universe of distressed value (high book-to-market) firms from current- and prior-year statements, prints the per-signal table, and reports the return spread between the high-$F$ and low-$F$ baskets - the exact screen Piotroski described.

The screen does exactly what Piotroski claimed it does. All six firms are “value” (high B/M), yet the four financially strong ones (F-score 7–9) average 17.00% while the two distressed ones (F-score ≤4) average 8.75% - an 8.25% value+quality spread within the cheap universe. BedrockOil and QuarryRail pass all nine signals (positive, improving, cash-backed earnings; falling leverage; improving liquidity); GraniteLumber and CastleIron are the distressed names the screen is built to exclude.

---

4. Failure Modes & First-Principles Breakdowns

1. The F-score rewards any improvement, even from a terrible base. A firm whose loss narrowed from −40% to −10% still scores F_ROA = 1 and F_AROA = 1. It is a distressed-value screen, not a universal quality score - Piotroski’s conditioning on high B/M is the point, and applying it to healthy growth firms produces ambiguous signals.
2. Binary signals discard magnitude. The F-score throws away how much a signal moved; two firms with very different degrees of improvement can score identically. It is deliberately coarse (robust to noise, cheap to compute), and that coarseness is a feature only inside its target universe.
3. Quality is as good as the inputs. Restated, smoothed, or fraudulent statements poison every signal - the accrual signal (CFO > ROA) exists precisely to catch earnings-management, but it catches detectable manipulation only (Accounting Quality & Red Flags).
4. Accruals vs. cash is a continuum, not a binary. Sloan’s factor is a magnitude signal; collapsing it to “CFO > ROA” loses the size of the accrual gap. For a precise quality factor, use the continuous accrual ratio, not just the sign (Chan et al. 2006).

---

5. References

• Piotroski, Joseph D.: “Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers” (JAR, 2000, 38 Supplement)
• Sloan, Richard: “Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings?” (TAR, 1996)
• Chan, Konan; Chan, Louis; Jegadeesh, Narasimhan & Lakonishok, Josef: “Earnings Quality and Stock Returns” (JF, 2006)
• Dechow, Ge & Schrand: “Understanding Earnings Quality” (JAE, 2010)
• Greenblatt, Joel: The Little Book That Beats the Market

---

6. Connected Graph Bridges

• Back: 03 · Value & Profitability · Index Hub
• Forward: 05 · Failure Modes · 06 · Advanced Extensions
• Defense: Accounting Quality & Red Flags · The Accrual Anomaly
• Base: Core Financial Ratios · Advanced Extensions