Trading Playbook
How to read signals, size positions, and exit. Distilled from Benjamin Graham (margin of safety), Joseph Piotroski (financial health), Edward Altman (distress), and our own observations on Indian markets.
The 6 signal types
See live signals →STRONG SETUP
Score ≥70 · 52W-pos ≤25% · Piotroski ≥7 · FCF ≥2% · MoS ≥20%
Highest conviction. Book 33% at +15%, 33% at fair value, trail 34%. All 5 conditions must hold — this is rare by design.
QUALITY SETUP
Score ≥65 · 52W-pos ≤40% · Piotroski ≥6 · FCF ≥1% · MoS ≥10%
Relaxed STRONG SETUP. 40% at +10%, 40% at fair value, trail 20%. Wider net catches good compounders missing one criterion.
COMPOUNDER
ROCE ≥20% improving · FCF >0 · Score ≥55
Buffett mode. No price target — hold for the compounding. Exit only on fundamental deterioration (ROCE below 15%, D/E above 0.5).
TURNAROUND
Piotroski ≥7 · Score ≥50 · 52W-pos ≤50%
Classic "fallen angel." Balance sheet is healing, price still beaten down. Give it room — turnarounds are volatile.
GRAHAM SETUP
Graham MoS ≥15% · EPS >0 · D/E <0.5
Pure Benjamin Graham defensive formula. Can drift sideways for 18+ months — Graham himself said the market is a voting machine in the short run.
WATCH
Score 50–64 · 52W-pos ≤35% · MoS ≥5%
DO NOT BUY. Research list only — use the time to understand the business so you can act fast when it upgrades to BUY.
Rules that override everything
⏲ Entry scaling
Never buy 100% at once. Week 1: 40% of intended size. Week 2–3: 30% more if thesis intact. Week 5–6: final 30% if price has not run away.
📦 Position sizing
Single stock cap: 10% of portfolio. Single sector cap: 30%. GRAHAM SETUP names (long drawdowns) capped at 3%. STRONG SETUP + TURNAROUND CONFIRMED combined: up to 8% each.
🚨 Override — do not buy even if signal fires
Promoter pledge >10% · Promoter hold dropped >5% last quarter · Altman Z <1.8 · FCF negative 3+ of last 5 years · You do not understand the business · Recent auditor change or accounting restatement.
💰 Book-profit discipline
Never sell 100% at once — you will regret it. Always book something at T1, even 25% of position. Let trailing stop handle T3; do not try to pick the top.
⏱ Time horizon is a filter
If you cannot hold through 2 market crashes, do not buy compounders. If you check the price every hour, size down. Value re-rating happens in quarters, not days.
📝 Write it down before buying
Three questions: Why is it cheap? (Temporary vs structural.) What would make me sell? (Fundamental trigger, not price.) Can I hold 18 months without checking daily? If any answer is unclear, skip.
The U-Score framework (0–100)
Every stock gets a 0–100 overall value-opportunity U-Score made of 5 pillars plus bonuses and penalties. It is not a pure valuation label.
- Valuation→30 pts
- Growth→25 pts
- Quality→20 pts
- Balance Sheet→15 pts
- Cash Flow→10 pts
- Piotroski bonus→5 pts
These names currently map the total U-Score, so an “Overvalued” legacy band does not mean that P/E alone is excessive. A separate valuation-evidence model and Value-trap risk are running in parallel shadow mode before any public replacement. Stocks with newer reported results are marked for recalculation, and the shadow release gate requires same-source quarter coverage rather than relying on a fetch timestamp alone.
Banks use a separate valuation track (Gordon Growth justified P/B) because their balance sheets are structurally leveraged. See packages/algorithm for the full implementation.