What is margin of safety?

Benjamin Graham coined the term, and the logic behind it is almost mechanical: if your estimate of what something is worth could reasonably be off by 20%, you don't want to pay a price that only makes sense if your estimate is exactly right. You want enough of a gap between price and estimated value that your estimate can be wrong and you still come out fine. That gap is the margin of safety.

How it's measured

There are a few conventions; the one we use (matching Alpha Spread) measures the gap against the larger of the two numbers:

• Undervalued = (fair value − price) ÷ fair value
• Overvalued = (price − fair value) ÷ price

A stock that's 50% undervalued on this measure is trading at half its estimated fair value. A stock that's 50% overvalued has a price that's twice its estimated fair value. The measure is symmetric and bounded between −100% and +100%, which keeps a single extreme outlier from distorting how you read it.

Why it matters more than the number itself

A fair value estimate is a single point drawn from a model full of assumptions — growth rates, margins, a discount rate, a terminal growth rate. Any of them could reasonably differ from what actually plays out. A large margin of safety is what makes that uncertainty survivable: if the true value turns out to be 15% lower than estimated, a stock bought at a 40% discount is still a reasonable purchase, while a stock bought at a 5% discount is now a mistake.

This is also why margin of safety isn't the same question as "is this a good company." A wonderful business bought at a price that already reflects its quality has no margin of safety at all — there's no room for the story to go even slightly wrong. A mediocre business bought cheaply enough can still offer one. The two questions — how good is this business, and how much am I paying for that — are genuinely separate, and margin of safety is squarely about the second one.

What it isn't

It isn't a guarantee. A large discount to a model's fair value can mean the market has spotted something real that the model hasn't — weak accounting controls, a melting competitive position, a customer concentration risk — in which case the "cheap" price is correctly cheap, not a bargain. Margin of safety reduces the damage from being wrong about the valuation; it doesn't eliminate the possibility that the valuation itself is missing something. Treat a large gap as a reason to look closer, not a reason to stop looking.