Mean vs Median Valuation Multiples

Which statistic represents this valuation peer set?

The task is to select a multiple for Example Company’s target EBITDA of 800 in EUR thousands. Start with Trading Comparables to define the peers and metric. This exercise keeps the target metric and peer observations fixed while testing the statistic used to apply EV / EBITDA.

Enter synthetic comparable-company observations

Peers A–E each have LTM EBITDA of 1,000. Supplied enterprise values are 6,000, 8,000, 10,000, 12,000 and 24,000, giving 6×, 8×, 10×, 12× and 24×. All five initially have Include selected. These are invented observations, not live prices or a selected real peer universe.

The target is 2026 reported EBITDA, 800, from the synthetic operating forecast in FCFF Calculation. Target EV is on an enterprise basis. The Enterprise Value vs Equity Value bridge is a separate step; do not treat the EV shown here as the equity purchase price.

Comparable-company statistics: reconcile mean, median and quartiles

The five included observations give mean 12×, median 10×, 25th percentile 8× and 75th percentile 12×. Each included company contributes one observation, not a market-capitalization weight. The 24× peer pulls the arithmetic mean upward; the middle sorted observation remains 10×.

Select Mean to obtain target EV 800 × 12 = 9,600. Select Median for 800 × 10 = 8,000. The difference of 1,600 follows from statistic selection alone. The quartile-based target range is 6,400–9,600; the product includes the selected multiple if it lies outside that quartile interval. This range is not a confidence interval.

Test an economically justified exclusion

For the exclusion case, assume Peer E has a different business model and growth profile that your comparability review finds unsuitable for this target. Clear only E’s Include flag. Its 24× value alone is not a sufficient reason to exclude it. The engine retains the observation but omits it from the included statistics.

The remaining 6×, 8×, 10× and 12× observations yield both mean and median of 9×. The implemented interpolated quartiles become 7.5× and 10.5×, not simply the two middle companies. Either central statistic now gives EV 7,200; the quartile range is 6,000–8,400. The target EBITDA remains 800.

Choose the method before looking for the desired value

Median may represent the central company better when a comparable but extreme observation stretches the mean. Mean may be useful when the dispersion itself reflects the relevant peer economics. Neither is universally preferable. In this example, excluding a genuinely different peer changes the economic sample and both central statistics, not merely the presentation.

Record why E belongs or does not belong and why the chosen statistic fits the target. Do not switch between mean and median solely to obtain a desired valuation. Precedent Transactions uses transaction observations with a different economic basis; Football Field Valuation compares the resulting method ranges.

Reproduce the supported statistic selection

1. Download the native trading-comps example, open Trading Comparables and explicitly use Manage models → Import. In Model, switch to Advanced before editing, then return to Value or Deals as appropriate. A blank workspace may first show Activate; the example file already has the relevant tool active. The CTA opens your workspace and never imports the example automatically.

2. In Trading Comparables keep target year 2026, reported basis, peer period LTM, metric EV / EBITDA and the selected-multiple override blank. The file starts at Median with all five peers included. Open peer observations to inspect EV and EBITDA, then read statistics and target EV.

3. Under Apply Trading Comps, change Statistic to Mean and check 12× / EV 9,600. Restore Median and check 10× / EV 8,000. Clear Peer E’s Include flag and compare mean, median and both quartiles; either central statistic gives EV 7,200. Restore E to return to the starting set.

What the product does not decide

The engine calculates statistics over supplied, included observations. It does not fetch market data, choose peers, detect outliers automatically or recommend mean versus median. Missing or nonpositive metric denominators do not create valid multiples. A direct multiple override bypasses the statistic-selected value, so leave it blank for this exercise.