2026-09-20 · Valuation

Trading Comps: A Workflow You Can Actually Reproduce

Comps — relative valuation, comparable company analysis — is the most used and most easily faked method in equity work. It's simple enough to fill in on one sheet, and precisely because it's simple most people stop at "find a few peers, paste some multiples, take an average." This piece compresses comps into an eight-step reproducible workflow, which is also the core of the valuation-comps skill.

First: what question does comps actually answer?

Comps does not answer "what is this company worth" in absolute terms. It answers: what multiple is the market currently willing to pay for this kind of business? It's a relative anchor, not intrinsic value. Admit that before you decide anything with it — multiple expansion and contraction explain short-term moves more often than fundamentals do.

The eight steps

1. Define the target: name the subject (ticker or company) and fix the vantage — buy-side (should I buy) or sell-side (what should it fetch).
2. Screen peers: filter on three axes — same business, comparable scale, same market — not on an industry label. Three to eight names is the sweet spot; fewer is noise, more dilutes the signal.
3. Pull multiples: EV/Revenue, EV/EBITDA, P/E, P/S as appropriate. Growth names lean on EV/Revenue; stable earners on EV/EBITDA and P/E.
4. Align definitions: use normalised multiples (NTM or FY+1), strip one-off items, unify capital-structure treatment.
5. Compute the centre: report the median as primary and the mean as secondary — the median is far steadier against outliers.
6. Run sensitivity: 25th / 50th / 75th percentile multiples against the target's forecast metric to produce a range, not a point.
7. Cross-check: reconcile the comps result against DCF and precedent transactions. A large gap means you go back and re-examine peer selection.
8. Conclude: state expensive / fair / cheap explicitly, and name the variables that would trigger a re-rating. Not a price target.

A sample table (illustrative figures, not real data)

The numbers below exist only to show the output shape:

If the target trades at 10x EV/EBITDA against a peer median of 14x, it looks cheap — but you have to account for the discount. Slower growth, more leverage, or a turnaround the market hasn't priced yet? Cheapness without a reason is a value trap.

Three mistakes that wreck most comps

• Peers too broad: forcing companies with different business models into one set makes the multiples meaningless.
• Mean only: one or two extreme valuations drag it. The median is the stable anchor.
• Treating comps as truth: multiples reflect sentiment, not intrinsic value. Relative and absolute valuation should corroborate each other, not replace each other.

This workflow is packaged as the valuation-comps skill — give it a target and it runs the eight steps and produces the same table. If it can't be re-run, it isn't worth building.

← All articles