Key points
- 01The original Piotroski F-score uses nine binary signals across profitability, cash quality, leverage, efficiency, and dilution within high book-to-market stocks
- 02F-score compresses past financial improvement rather than proving absolute business quality, so industry and life-cycle context remains necessary
- 03COMP reads market expectations and relative strength while F-score reads accounting resilience; agreement and conflict between them can guide research priority
- 04Valuation is better expressed as ranges and sensitivity across DCF, relative, and asset approaches than as one precise target price
- 05Price stops, thesis breaks, and value realization solve different problems, and a stop order does not guarantee execution at the stated stop price
Chapter 1. Use financial statements to separate recovery from deterioration
From momentum's limits to F-score
Price momentum can capture gradual information diffusion but cannot diagnose balance-sheet deterioration. Joseph Piotroski designed F-score to separate financially strong and weak firms inside a high book-to-market universe. The original research reported an improved return distribution in its 1976–1996 sample, not a universal promise for every current market.
Five interpretive groups behind nine signals
The original score assigns one point each for positive net income, positive operating cash flow, improving return on assets, accrual quality, falling leverage, improving current ratio, no new equity issuance, improving gross margin, and improving asset turnover.
For interpretation, those nine signals can be grouped into five questions: profitability, conversion of earnings to cash, leverage and liquidity, margin and asset efficiency, and dependence on equity issuance. This grouping does not replace the original specification; it reveals how two equal totals can represent different risks.
Beyond binary F-score and a five-step quarterly routine
Continuous industry percentiles for cash flow to assets, leverage change, and margin change can restore magnitude, but every new feature raises overlap and overfitting risk. Each metric should have a defined job before it is added.
The quarterly routine is: record the actual release date; normalize one-offs and accounting changes; calculate binary and continuous signals; compare results with the prior thesis; then choose add, hold, reduce, or reject and write the next falsification condition.
Chapter 2. View the market through two lenses
When COMP and F-score point to the same company
A high COMP rank suggests that price and expectations are improving relative to peers. A high F-score suggests that recent accounts support financial strength. Their agreement is useful evidence, but the two may still reflect the same event at different reporting lags.
Reading the two ranking tables
Read COMP components before the total: momentum with falling quality is different from balanced strength. Rank changes can occur because peers improve even when the company does not, so raw changes must remain visible.
F-score has many ties because it is an integer from zero to nine. Cash-flow magnitude, debt maturity, margin change, audit opinion, and working capital can restore context. Financial institutions and early-stage growth firms often require separate models because leverage and cash investment have different meanings.
Candidates that pass both lenses
High COMP and high F-score names move to deeper research. High COMP with weak F-score asks whether price has outrun financial improvement. Weak COMP with strong F-score may be healthy but lack a catalyst. Quantitative rankings are most useful for ordering research rather than automating a purchase.
Chapter 3. What price to pay: valuation and targets
Valuation is a system of assumptions
Value compares uncertain future cash flows with today's price. Enterprise value, equity value, free cash flow, discount rate, terminal growth, and diluted share count are the inputs that turn a narrative into an explicit model.
Market, DCF, and asset approaches
Relative valuation uses peer multiples and reflects the market's current language, but fails when peers are not comparable or an entire sector is mispriced. DCF makes growth, margin, reinvestment, and risk assumptions explicit, but a distant terminal value can dominate the answer. Asset value is useful for financial, property, resource, or liquidation cases, while often understating intangible growth capacity.
Different industries require different economic anchors: capital adequacy and ROE for banks, recurring revenue and cash conversion for software, normalized cycle earnings and capital expenditure for manufacturing, and commodity scenarios for resource firms.
Cross-validating a target range
Damodaran notes that DCF and relative valuation can disagree because they make different assumptions about where the market is wrong. The task is not to average the numbers automatically, but to explain the difference and identify overlapping confidence ranges under bearish, base, and bullish scenarios.
Chapter 4. Integrating sell rules: deciding when to exit
Different rules solve different problems
A price rule protects the portfolio, a fundamental rule reflects a broken thesis, and a valuation rule reallocates capital after expected return falls. None fully replaces the others.
A stop order may become a market order after it is triggered, producing an execution far from the stop in a fast market. A stop-limit controls price but may not execute at all.
A three-stage integrated logic
Stage one is immediate survival control after fraud, funding failure, or a regulatory event invalidates the thesis. Stage two is a thesis review when estimate cuts, cash-conversion weakness, and relative-strength breakdown overlap. Stage three is capital reallocation when price exceeds the upper valuation range and expected return falls below available alternatives after tax and costs.
Re-entry should also be a new decision under written rules. A prior sale does not prevent repurchase after the thesis genuinely recovers and both market and financial signals again pass the required thresholds.
Two scores and three valuation methods can still share one wrong assumption
Cross-validation is strongest when independent evidence agrees. Price momentum, analyst estimates, and management guidance may all depend on the same optimistic narrative, making apparent confirmation less independent than it looks.
Financial scores describe accounting history, valuation describes future assumptions, and price describes current market expectations. Their dates and source lineages should be recorded separately.
The integrated sell process is intended to order decisions, not to promise automatic protection. Execution risk, liquidity, tax, and personal time horizon remain separate constraints.
Sources
Related notes
Direction of MoneyInvesting in global market leadersApplies one factor discipline across markets while incorporating currency, disclosure, liquidity, and concentration into selection and allocation
Research noteBuffett, Druckenmiller, and Bridgewater require three different readings of a 13FReading concentrated compounding, tactical rotation, and systematic diversification in the second-quarter 2026 filings
Direction of MoneyReading the market through an investor's eyesExplains how a few long-run winners create most market wealth and how value, growth, momentum, and quality can frame a search for future leadersThis is a personal research note, not investment advice