Interpreting Comps Output: Mean, Median, Percentiles, and Outlier Handling

Introduction

Once the comps table is built, with peer companies identified, financial data normalized, and multiples calculated, the analyst must interpret the output. The raw data (a column of EV/EBITDA multiples ranging from, say, 7.5x to 18.2x) does not automatically produce a valuation. The analyst must decide which statistical measure best represents the peer group's pricing, how to handle companies that trade at dramatically different multiples, and how to translate the summary statistics into a defensible implied valuation range.

This interpretive step is where comparable company analysis transitions from data collection to analytical judgment. Two analysts with the same comps table can arrive at different valuation conclusions based on how they interpret the output.

The Median: Default for Most Peer Groups

The median (the middle value when multiples are ranked from lowest to highest) is the standard anchor in investment banking comps analysis. Its primary advantage is resistance to outliers. If a peer group of eight companies has multiples of 8x, 9x, 10x, 10.5x, 11x, 11.5x, 12x, and 25x, the median (10.75x) reflects the central tendency of the group, while the mean (12.1x) is pulled upward by the single outlier at 25x.

In most comps analyses with 5 or more comparable companies, the median is the appropriate anchor because:

The Mean: For Small Peer Groups

The mean (arithmetic average) is preferred when the peer group is small (fewer than 5 companies) and contains no clear outliers. With only 3-4 comparables, the median can be dominated by a single company, making the mean a more representative central measure. However, if even one of those 3-4 companies has an anomalous multiple, the mean becomes unreliable as well, which is one reason banks generally prefer peer groups of at least 5 companies.

Percentiles divide the peer group's multiples into segments. The most commonly used in investment banking:

The interquartile range (Q1 to Q3) is the most common range used for the comps bar on the football field chart. It captures the middle 50% of the peer group, excluding the most extreme values on both ends.

Interquartile Range (IQR)

The range between the 25th percentile (Q1) and 75th percentile (Q3) of a data set. In comps analysis, the IQR of the peer group's multiples defines the implied valuation range most commonly shown on the football field chart. The IQR is preferred over the full range (high to low) because it excludes extreme values on both ends, providing a more representative bracket of where comparable companies trade. A narrow IQR signals that peers are valued similarly by the market, increasing confidence in the benchmark. A wide IQR suggests greater dispersion, which may indicate that the peer group is less homogeneous or that the market differentiates significantly based on company-specific factors.

Outliers are peer companies whose multiples differ significantly from the rest of the group. They are common and require thoughtful handling, not reflexive exclusion.

Investigate Before Excluding

When a peer trades at a dramatically different multiple, the analyst's first task is to understand why:

Decision Framework

Raw summary statistics tell you the central tendency and range, but they do not explain why multiples differ across the peer group. The best analysts go beyond the statistics to build a narrative that connects multiple differences to fundamental differences:

A peer trading at 14x NTM EBITDA versus a group median of 10.5x might be justified by superior growth (20% revenue growth vs. the median's 8%), higher margins (35% EBITDA margin vs. 22%), or lower risk (more diversified revenue, less customer concentration). Conversely, a peer at 7.5x might reflect cyclical headwinds, margin pressure, or market-specific concerns.

This contextualization is what allows the analyst to assess where the target should fall within the range. If the target has growth and margins that exceed the peer median, it may warrant a multiple above the median. If it has below-median profitability or higher risk, a discount may be appropriate.

In practice, this assessment often takes the form of a scatter plot or a simple ranking. The analyst places each peer on a chart with revenue growth on one axis and EV/EBITDA on the other, confirming visually whether the expected positive correlation holds. If it does, the target can be positioned on the same chart based on its growth rate, and the implied multiple read from the trendline. Even without a formal regression, this visual exercise quickly reveals which peers the target most closely resembles and whether the median multiple or a higher/lower point is more appropriate. The best comps analyses do not just present a range; they explain why the target belongs at a specific point within it.

NM (Not Meaningful)

A designation used in comps tables when a calculated multiple produces a result that is not analytically useful. This occurs most commonly when EBITDA is negative (producing a negative EV/EBITDA multiple), when EBITDA is near zero (producing an extremely high multiple), or when the company is undergoing a transformational event that makes its current financials unrepresentative. Companies marked "NM" are typically excluded from the summary statistics (mean, median, percentiles) to prevent them from distorting the peer group benchmarks.