Benjamin Graham Margin of Safety Model

Benjamin Graham model inputs

Score each input from 0 to 100, multiply by the weight, then read the total with the score band table.

Benjamin Graham score bands

Use the score as a research gate, not as a price target or recommendation.

Benjamin Graham action matrix

The action matrix converts model conditions into the next research step.

Investor Checklist

• Balance sheet safety (25%): Net cash, debt maturity, interest coverage, liquidity, refinancing risk.
• Normalized earnings (20%): Average-cycle earnings, margin floor, cyclicality, one-time gains or losses.
• Cash conversion (15%): Free cash flow conversion, capex needs, working-capital pressure.
• Valuation discount (20%): Price versus conservative earnings, book value, replacement value, and peer value.
• Downside protection (15%): What the business is worth if growth slows, margins compress, or estimates fall.
• Catalyst patience (5%): Whether the discount can close without forcing a short-term trade.

The Model

Graham Margin of Safety Score: Score = 25% balance sheet safety + 20% normalized earnings + 15% cash conversion + 20% valuation discount + 15% downside protection + 5% catalyst patience.

• Balance sheet safety (25%) — Net cash, debt maturity, interest coverage, liquidity, refinancing risk.
• Normalized earnings (20%) — Average-cycle earnings, margin floor, cyclicality, one-time gains or losses.
• Cash conversion (15%) — Free cash flow conversion, capex needs, working-capital pressure.
• Valuation discount (20%) — Price versus conservative earnings, book value, replacement value, and peer value.
• Downside protection (15%) — What the business is worth if growth slows, margins compress, or estimates fall.
• Catalyst patience (5%) — Whether the discount can close without forcing a short-term trade.

Scorecard and action tables

Use the tables above as the working model. Fill each input with current evidence, assign a rough score, then let the action matrix decide whether the page deserves deeper research.

• Cheap + strong balance sheet + normalized FCF: Continue research — Compare bear/base/bull forecasts and check 13F value-manager context.
• Cheap + high debt: Wait — Debt can erase the margin of safety before valuation repairs.
• Cheap + peak margins: Normalize first — Do not score the stock on unusually strong cycle earnings.
• Not cheap + high quality: Use another model — Graham may under-score asset-light compounders.

How To Read The Score

The score is not a price target. It is a research gate. High scores mean the setup deserves deeper work; low scores mean the model is warning that the thesis is incomplete.

• 85-100: Research candidate — Cheap, solvent, cash-generative, and downside case remains acceptable.
• 70-84: Watchlist only — Some safety exists, but valuation, debt, or earnings normalization still needs confirmation.
• 50-69: Value trap risk — Looks cheap, but one or two core safety variables are weak.
• 0-49: Reject — Downside is not protected enough to justify the discount.

Hard Invalidation Rules

These rules override the score. If one hard invalidation appears, the model should be re-scored before any position decision.

• Debt refinancing becomes harder or more expensive.
• Free cash flow no longer tracks reported earnings.
• The bear case implies dilution, covenant pressure, or permanent impairment.
• The discount exists because the business is structurally shrinking, not temporarily unpopular.

Example Model Run

A cyclical industrial stock trading at 9x peak earnings might score only 58 if normalized margins are much lower and debt is high. The model would label it value-trap risk, not a bargain.

SnowballHare Data Workflow

Run the model with SnowballHare pages as evidence sources. The goal is to make each score traceable, not subjective.

• Use stock forecasts to check current valuation, earnings estimates, and scenario ranges.
• Use earnings analysis to normalize margins and cash flow after a noisy quarter.
• Use 13F investor pages to see whether value-oriented managers hold or are changing exposure.
• Use compare pages to avoid buying the statistically cheapest peer when another company has a stronger balance sheet.

Model Traps

The model becomes dangerous when one input is treated as the whole answer.

• Treating a low P/E as proof of value without checking earnings quality.
• Ignoring leverage because the stock looks cheap.
• Using last year's peak earnings as normal earnings.
• Applying asset-value logic to a business whose value is mostly intangible.

When To Use Another Model

It can underweight intangible assets, network effects, software margins, brand power, and long-duration growth. A business can look expensive on Graham metrics while still being a high-quality compounder.

Related Topics

Common Questions