First define the discount denominator
If estimated value is 100 monetary units and price is 75, the discount to value is 25%. The possible rise from 75 to 100 is about 33.33%. These are different ratios: the first denominator is 100 and the second is 75. The larger percentage is not evidence of a larger margin; the basis changed, not the economics of the offer.
Also separate the difference before and after costs. Transaction price, commissions and mandatory related costs may differ from a quotation. If the example uses a price before costs, state that limitation beside it. Applicable taxes are not specified here; they require a separate check of period, jurisdiction and circumstances.
Cheaper than the central estimate, more expensive than the stress case
Suppose a central estimate is 100, price is 75 and an agreed adverse scenario gives 60. The difference from the central calculation is positive, but relative to the stress case the price is 25% higher. If the scenario occurs and the market values the asset at 60, the decline from the purchase price is 20% before costs.
This example assigns no probability of loss and does not make 60 a lower bound. It shows why one discount to a central number guarantees nothing. A stronger deterioration may occur outside the considered events, and the central estimate may be systematically too high. Guaranteed income cannot be inferred from hypothetical figures; the SEC separately warns about the limits of such claims.[27](#source-market-performance)
Errors that a small discount does not fix
A missing obligation, an incorrect shareholder percentage or a need for additional financing can change an estimate abruptly. A slightly lower growth rate is not a full check of these errors. Before discussing margin of safety, reconstruct the rights to cash flow and confirm that the relevant share really belongs to the owner of the security being analysed.
Another problem is dependent assumptions. If high margins, cheap financing and rapid growth all rely on one favourable market condition, they are not three independent confirmations. A discount to a model built on such assumptions may not compensate for the common error in the underlying story.
Observable reasons to revise
Define in advance what information will change the value: loss of a major customer, unexpected capital intensity, a change in share count or lower returns on new projects. Price alone does not explain what happened to the business. A fall can increase the gap from the previous calculation while new information simultaneously lowers the estimate itself.
The review must not become endless fitting. Preserve the previous model version, date and reason for each change. If every deterioration merely moves the date for reaching the former profit one year later, ask whether there is observable support for recovery. A change history helps identify when “patience” has replaced fact checking.
The final memo should show several estimates, price with stated costs, key errors and review conditions. It may conclude “margin not confirmed” or “insufficient data”. It is not personal advice to enter or avoid a transaction. Margin of safety is a way to discuss uncertainty in price and valuation, not capital insurance, a recovery guarantee or a promised return.
A discount does not compensate for lack of understanding
When the underlying economics are unclear, increasing the required discount does not always solve the problem. If obligations or ownership of the cash flow cannot be assessed, even a very low price relative to the selected model may have uncertain meaning. Distinguish a wide but meaningful range from a situation where the valuation object itself has not been reconstructed.
Check the waiting period separately. The difference between price and model value does not promise that the market will recognise it by a particular date. If obligations require an exit earlier than the thesis assumes, the observed discount does not remove calendar risk. Waiting cannot be declared free merely because the model has no separate time commission.
The conclusion should therefore include not only discount arithmetic but also what remains fundamentally unknown. Sometimes the most disciplined conclusion is not to refine the percentage, but to acknowledge the limits of available knowledge.
Related foundations: [Discounted cash flows: a model with open assumptions](/guides/discounted-cash-flow/); [Growth and value: two languages of expectations, not two lists of winners](/guides/growth-and-value/).