How I score companies
These thresholds were fixed before I researched a single company, and they apply to every one of them. If you want to check a score anywhere on this site, this is the page that lets you.
A score you cannot reproduce is an opinion with a number painted on it. Publishing the rulebook lets you verify not just what I concluded, but that the method was fixed in advance and that I cannot quietly move a threshold to flatter a company I like. Every published score also stores the rulebook that produced it, so later revisions never rewrite history.
Operating companies
| Metric | 5 | 4 | 3 | 2 | Direction | Bucket |
|---|---|---|---|---|---|---|
| Avg ROCE, 5 years | ≥20% | ≥15% | ≥12% | ≥8% | higher better | Quality |
| Avg ROE, 5 years | ≥18% | ≥14% | ≥11% | ≥8% | higher better | Quality |
| Avg operating margin | ≥20% | ≥15% | ≥10% | ≥6% | higher better | Quality |
| Margin trend | ≥2pp | ≥0pp | ≥-2pp | ≥-5pp | higher better | Quality |
| Cash conversion | ≥0.9 | ≥0.75 | ≥0.6 | ≥0.4 | higher better | Quality |
| Revenue CAGR, 5 years | ≥15% | ≥10% | ≥6% | ≥3% | higher better | Growth |
| PAT CAGR, 5 years | ≥18% | ≥12% | ≥7% | ≥3% | higher better | Growth |
| P/E vs own 5y average | ≤-20% | ≤-5% | ≤10% | ≤30% | lower better | Valuation |
| P/E vs peer average | ≤-15% | ≤0% | ≤15% | ≤35% | lower better | Valuation |
| Debt / equity | ≤0.1 | ≤0.3 | ≤0.6 | ≤1 | lower better | Risk |
| Interest cover | ≥10 | ≥6 | ≥3.5 | ≥2 | higher better | Risk |
Banks and lenders
| Metric | 5 | 4 | 3 | 2 | Direction | Bucket |
|---|---|---|---|---|---|---|
| Avg NIM, 5 years | ≥4% | ≥3.5% | ≥3% | ≥2.5% | higher better | Quality |
| Avg ROA, 5 years | ≥1.8% | ≥1.4% | ≥1% | ≥0.6% | higher better | Quality |
| Avg ROE, 5 years | ≥17% | ≥14% | ≥11% | ≥8% | higher better | Quality |
| Avg CASA ratio | ≥45% | ≥38% | ≥30% | ≥22% | higher better | Quality |
| Advances CAGR, 5 years | ≥18% | ≥14% | ≥10% | ≥6% | higher better | Growth |
| PAT CAGR, 5 years | ≥18% | ≥12% | ≥7% | ≥3% | higher better | Growth |
| P/B vs own 5y average | ≤-20% | ≤-5% | ≤10% | ≤30% | lower better | Valuation |
| P/B vs peer average | ≤-15% | ≤0% | ≤15% | ≤35% | lower better | Valuation |
| Gross NPA | ≤1% | ≤2% | ≤3.5% | ≤5% | lower better | Risk |
| Gross NPA trend | ≤-0.5pp | ≤0pp | ≤0.5pp | ≤1.5pp | lower better | Risk |
| Capital adequacy | ≥18% | ≥16% | ≥14% | ≥11.5% | higher better | Risk |
Weights
| Bucket | Quality | Bank |
|---|---|---|
| Quality | 35% | 30% |
| Growth | 15% | 15% |
| Valuation | 20% | 15% |
| Momentum | 5% | 5% |
| Risk | 25% | 35% |
What the score decides
| Weighted score | Verdict |
|---|---|
| ≥ 4.00 | ADD CANDIDATE |
| 3.25 to 3.99 | WATCH / STARTER |
| below 3.25 | PASS |
These thresholds are fixed before a company is researched and applied to every company identically. Each published snapshot stores the rulebook that produced it, so an old score can always be checked against the rules in force at the time.
How a company moves through it
THE SEQUENCE| Step | What happens | Who decides |
|---|---|---|
| 1 | Figures are taken from Screener and the annual report | The source |
| 2 | Each figure is graded against a threshold fixed in advance | The rulebook |
| 3 | Moat, growth runway, momentum and governance are judged 1 to 5 | Me, and marked as mine |
| 4 | Five governance questions run as a hard gate | The rulebook |
| 5 | Buckets are averaged, weighted, and a verdict falls out | Arithmetic |
| 6 | The score and the rulebook that produced it are frozen on publication | Nobody, afterwards |
Nothing. If any bucket is missing or the governance gate is unanswered, the verdict reads INCOMPLETE and no score is shown at all. A partial reading cannot flatter itself into a number.
Where this model breaks, and I would rather say so
The thresholds assume a normal operating company or a lender. They do not fit insurers, REITs or holding companies well, and they do not apply at all to the instruments in Balanced Allocation, which is why that product is judged on allocation discipline instead. Momentum is a weak input and is weighted accordingly. And a rigorous score built on careless data is still careless: the rules discipline my judgement, they do not replace it.