Value Investment Checklist

Value-Investment Checklist

Prompt-first skill. Score the 30 criteria by reasoning over fundamentals. scripts/checklist.py is OPTIONAL. Score only criteria you can evaluate; surface the rest as missing:<field> and don't count them.

Role & objective

You run a 30-point value checklist across four schools — Buffett, Lynch (GARP), Graham, and Quality — and return a weighted Investment Grade (A+..F) with a 0–4.0 GPA.

When to use

"Is this a good long-term investment?", "does GP pass the value checklist?", "Buffett/Graham quality check", "margin of safety". Pair with fundamental-analysis for the fair-value range.

Inputs you need

Gather via dse-data-acquisition → fundamentals (roe, debt_to_equity, profit_margin, free_cash_flow, eps_history, pe, pb, peg, earnings_growth, current_ratio, dividend_yield, institutional_ownership, intrinsic/fair value, etc.). Optional macro.inflation (Quality check).

Method — score each criterion pass/fail, grouped in 4 weighted buckets

Mark a criterion ? (not counted) when its field is missing.

Buffett (weight 0.35): economic moat; ROE > 15%; Debt/Equity < 0.5; profit margin > 20%; free cash flow positive; quality management (ROE>15% & margin>15%); predictable earnings (all EPS history positive); ≥25% below intrinsic value (margin of safety); business understandable; sustainable competitive advantage.

Lynch / GARP (weight 0.30): PEG < 1.0; earnings growth 15–30%; revenue growth consistent with earnings (±10pp); inventory turnover improving; insider buying; institutional ownership < 60%; share buyback in place; P/E < earnings-growth rate.

Graham (weight 0.35): P/E < 15; P/B < 1.5; P/E × P/B < 22.5 (Graham number); current ratio > 2; pays a dividend; (plus the remaining classic safety/earnings-stability checks).

Quality (weight ≈0.333): the four quality criteria, including growth beats inflation (uses macro.inflation, default 6%).

Scoring rubric

Each bucket score = (criteria passed)/(criteria evaluable). The four bucket fractions are weight-averaged (Buffett .35, Lynch .30, Graham .35, Quality ≈.333) and renormalised to sum to 1.0 → overall 0..1 score.

Grade & GPA: ≥0.90 A+ (4.0) · ≥0.80 A (3.7) · ≥0.70 B+ (3.3) · ≥0.60 B (3.0) · ≥0.50 C (2.0) · ≥0.40 D (1.0) · else F (0.0).

Confidence rises with the fraction of criteria evaluable; lower it when many fields are missing.

Output (emit this Thinking Card)

{ "skill": "value-investment-checklist", "ticker": "..", "mode": "investment", "as_of": "..",
  "score": 0.0, "confidence": 0.0, "rating": "B+",
  "key_metrics": { "gpa": 3.3, "overall_count": "10/30", "criteria_passed": 0,
    "criteria_evaluated": 0, "buckets": { "buffett": {"criteria_met": 0, "total": 0, "fraction": 0.0} } },
  "reasoning": ["✓/✗/? criterion — beginner-friendly why"], "flags": ["missing:<field>"],
  "disclaimer": "Educational analysis only. Not financial advice." }

DSE pitfalls

• DSE disclosure is patchy — many criteria will be ? (not counted); grade on what's evaluable and surface missing:<field> rather than failing unknowns.
• US-calibrated thresholds (P/E<15, margin>20%) are strict for DSE sectors — report the actual values so the user can judge in local context.
• A high grade on 8/30 evaluable criteria is low-confidence — say so.

Optional precision helper

python3 scripts/checklist.py --input data.json --pretty

Returns each criterion's pass/fail with its actual value, the bucket fractions, GPA and grade.

Worked example

LHB (partial data): Buffett 6/7, Lynch 1/2, Graham 3/4, Quality 0/0 evaluable → weighted ≈ 0.71 → B+, GPA 3.3, with several ? criteria flagged for missing fields.

References

Full 30 criteria & thresholds: references/CRITERIA.md. Output is educational analysis only, never financial advice.