Margin of safety, and what it costs: known limitation 1

Applying a margin to a valuation that was already conservative. If the rent assumption is cautious, the growth assumption is cautious and the exit price is cautious, then a further discount on the result counts the same risk two or three times and rejects almost everything for reasons that were already in the number.

Margin of safety, and what it costs: known limitation 2

Treating the margin as a substitute for doing the valuation properly. A percentage taken off an incomplete or biased estimate is a variation on a figure that was close to arbitrary, and it provides protection only in the sense that any refusal to act does.

Margin of safety, and what it costs: known limitation 3

Using one figure for every kind of asset. A margin that suits a tenanted apartment in a mature building with three years of accounts is far too small for a tower that does not exist yet, and applying the same number to both is a decision not to think about which is which.

Playbooks / Valuation

Margin of safety, and what it costs

Buying only at a discount to your own estimate of value protects against the estimate being wrong, and the discount is not free, because every point of it also turns away things that were worth buying.

The rule

Every valuation is an estimate, and every estimate is wrong by some amount in some direction. The margin of safety is the oldest answer to that: work out what a thing is worth, then refuse to pay anything close to it.

It is Graham's idea, it is genuinely useful, and it is more often invoked than applied. The version that gets quoted is buy at a discount. The version that survives scrutiny has a cost attached, and the cost is where the thinking is.

The rule, stated plainly

Estimate value. Apply a discount. Buy only below the result.

A twenty five percent margin on a value of one point six million means buying below one point two million. Not at one point five, on the grounds that it is still below the estimate. Below the discounted number or not at all, because the entire function of the margin is to be a line you do not cross when a specific unit is in front of you and you want it.

What it actually buys, and what it costs

This is the part usually left out.

A margin of safety trades one kind of error for another. Widening it reduces the chance of buying something that turns out to be overvalued. It also increases the chance of standing aside from something that was genuinely cheap. Those move in opposite directions, and no setting reduces both.

Damodaran's framing of this is the most useful one in print: the margin has a price, and the price is paid in investments not made. Somebody who has never once regretted passing on a purchase has probably set the margin so wide that it is no longer a discipline, it is a reason never to act.

The honest position is that the discount is a choice about which error you would rather make, and that a person who cannot name the cost of their own margin has not understood it.

The three ways it is misused

Double counting. If the rent assumption was already cautious, the growth assumption was already cautious and the exit price was already cautious, a further discount on the output counts the same risk twice. The value estimate should be an honest expectation, and the margin should sit on top of it, visible. A conservative valuation with a margin bolted on rejects nearly everything, and rejects it for reasons that were already in the number.

Rescuing a bad valuation. A discount applied to a figure that was incomplete or biased is a percentage of something close to arbitrary. Doing the discounted cash flow properly comes first; the margin protects against the residual uncertainty in a decent estimate, not against not having done the work.

One number for everything. A tenanted apartment in a building with three years of accounts and a verified service charge is a different estimation problem from a tower that does not exist yet. The same margin cannot be right for both, and applying it to both is a decision not to notice the difference.

Why property needs a wider one than shares

Three things widen the correct margin, and property has all three.

The estimate is shakier. There is no daily price, comparable sales are few, and the price per square foot inferred from three transactions in a quiet quarter is a number about three buildings.

The position is larger. A single apartment is often the biggest thing on the balance sheet, so being wrong about it is not diversifiable in the way being wrong about one holding in a global fund is.

The exit is slow and expensive. Eight to ten percent round trip and months of marketing means a mistake cannot be corrected cheaply, which is transaction cost drag and liquidity risk arriving together.

Off-plan compounds all three, which is the argument for treating the margin on a purchase from a plan as a different number entirely rather than the same one applied to a bigger figure. Off-plan payment plans has the arithmetic; this is the discipline that decides what to do with it.

How it relates to the other habit

Reverse the assumption asks what the future would have to do for the asking price to make sense. The margin of safety asks how wrong your own answer might be.

They are complements, and neither replaces the other. One interrogates the seller's number. The other interrogates yours.

The arithmetic

The rule Buy only if Price < Value x (1 - MOS) MOS of 25 percent on a value of 1,600,000 means buy below 1,200,000 and not above. The two errors it trades between Type 1 buying something overvalued Type 2 refusing something that was cheap A larger margin reduces the first and increases the second. There is no setting that reduces both. What it is not Not a substitute for a decent valuation. A percentage off a bad number is a bad number. Not a second risk adjustment. A conservative valuation with a margin on top counts the same risk twice. Not one number for everything. The same margin cannot suit a let apartment and an off-plan tower. Sizing it honestly Wider when the estimate is shakier, the position is larger, and the exit is slower. Property is all three.

Where it breaks

Applying a margin to a valuation that was already conservative. If the rent assumption is cautious, the growth assumption is cautious and the exit price is cautious, then a further discount on the result counts the same risk two or three times and rejects almost everything for reasons that were already in the number.
Treating the margin as a substitute for doing the valuation properly. A percentage taken off an incomplete or biased estimate is a variation on a figure that was close to arbitrary, and it provides protection only in the sense that any refusal to act does.
Using one figure for every kind of asset. A margin that suits a tenanted apartment in a mature building with three years of accounts is far too small for a tower that does not exist yet, and applying the same number to both is a decision not to think about which is which.
Forgetting that the margin has a cost measured in things not bought. Every increase in it reduces the chance of overpaying and increases the chance of standing aside from something that was genuinely cheap, and somebody who has never regretted a purchase they did not make has probably set it far too wide.
Assuming an individual margin gives portfolio safety. A large margin on every holding pushes a portfolio towards assets whose specific risks looked small, which is not the same as a portfolio whose risks offset, and diversification does work a concentrated margin cannot.

When to use it

Before any purchase where the value was estimated rather than observed, which in property is all of them. It belongs at the end of a valuation, applied deliberately and stated out loud, rather than smuggled into the assumptions along the way.

Sources

Last reviewed 28 August 2026. Educational research, not personal advice. Disclosure standards.