Start with the object being valued
Specify what is being valued: operating assets, the equity of the whole group or a particular share class. Date, currency and share count must match across alternatives. If one estimate is EV and another is equity value, their bounds cannot be combined into one corridor. First reconcile them to one object.
Also check what the market price means. A small last trade does not guarantee that a large block can be sold at the same price. A control valuation and the price of a small non-controlling interest may answer different questions. Do not add a control premium automatically: explain which additional actions and cash flows are actually available to the valued owner.
Three scenarios and the temptation of the average
Suppose a fictional company has equity values of 240, 300 and 390 million roubles with ten million identical shares. Per share, these are 24, 30 and 39 roubles. The arithmetic average is 31 roubles. It is not the “most likely value” unless the scenarios have justified probabilities.
Even if an author assigns subjective weights of 20%, 50% and 30%, the weighted result is 31.5 roubles. That is a mathematical consequence of the selected weights, not new empirical knowledge. The reader should see where the probabilities came from and understand that outcomes outside the three scenarios remain possible. A convenient table does not make the scenario set complete.
Why several methods are not independent confirmation
DCF and relative valuation may rely on the same normalized earnings and the same growth history. If the starting earnings are overstated, both methods can confirm one error. Agreement is useful only after shared dependencies are examined. Mechanically averaging two high valuations does not remove their common source of bias.
Disagreement should not be hidden either. It may reflect an expensive peer group, different risk assumptions or an incorrect perimeter. Show which variable explains most of the gap. Sometimes the conclusion is that a method is not applicable with the available data, not that another multiple must be chosen until the figures match.
How to label an honest range
Call it a “range of selected scenarios” and list the conditions at its bounds. Do not say “with 95% probability” unless an appropriate probability model has been built and its assumptions checked. Historical multiple dispersion is also not a probability corridor for future value. The SEC notes that hypothetical figures and historical results do not predict future performance.[27](#source-market-performance)
Round in proportion to data quality. If cash-flow uncertainty is measured in millions, kopecks per share create false precision. Keep the full calculations inside the file: presentation rounding must not obstruct reproducibility. State separately what new information could narrow the range and what events would make it obsolete.
The final card should contain the valuation object and date, scenarios, common error sources, the current observable price and applicability limits. It must not replace price with value or promise that the market will return to a central estimate. The difference is the beginning of a research question, not a ready recommendation. Openly expressed uncertainty is more useful than an unsupported exact number.
Update the range when the object changes
After an issuance, a major acquisition or a divestiture, the old range cannot be transferred to the new share without reconciliation. Even if the price remains inside the old corridor, rights and cash flows have changed. Preserve the valuation date and list events that require a new version.
When communicating with readers, separate the author’s “central scenario” from the “market expectation”. The latter cannot be reliably inferred from one quote without a model of the assumptions embedded in it. Different combinations of growth, risk and investment can produce one price. Calling the selected case the sole market forecast would be too strong.
Finally, explain the meaning of rounded bounds. Convenient round values improve clarity; they are not trading levels. A price crossing a scenario boundary does not by itself define a transaction: first check whether the underlying information and assumptions still hold.