Why Peer Comps Break
The arithmetic is nearly always right. It is the basis that quietly breaks, and most decks never disclose it.
A peer comp is the most trusted slide in finance, and the most quietly broken. Put five companies side by side, rank them on a multiple or a margin, and the ranking feels like arithmetic - objective, checkable, done. Most of the time the arithmetic is fine. The basis underneath it is not.
Every figure on that slide has already passed through a dozen silent decisions about what each line means. Change a few of them and the ranking reshuffles - not because anyone miscalculated, but because the companies were never being read the same way.
The arithmetic is the easy part
Dividing one number by another is rarely where comps go wrong. The division is trivial and the formula is auditable. What is not auditable, from the outside, is the numerator and denominator: what counts as debt, what counts as operating, what was treated as one-off, which period the figure even belongs to. Get those right and a simple ratio is sound. Get them inconsistent across names and the cleanest formula produces a confident, wrong ranking.
Where the basis quietly breaks
The fractures are always in the same places. The same line item is classified differently from one company to the next - leases capitalised here and expensed there, an item above the operating line in one filing and below it in another. One-offs are stripped out by one analyst and left in by another. And calendarisation - aligning mismatched fiscal years - silently shifts which twelve months you are even comparing.
None of these is an error. Each is a defensible reading. But a comp assumes every company on the page was read on one basis, and almost none disclose whether that is true.
Why the deck never shows it
Because the normalization happened privately, in someone's model, before the slide existed. The assumptions don't travel with the numbers, so the reader sees a tidy ranking and no trace of the choices that produced it. When two desks disagree, they argue about the conclusion - never the basis, because neither basis was ever written down.
Comparable by default
The fix is not a better formula; it is a shared basis. Standardize every company onto one consistent set of definitions, make each mapping visible, and let every figure trace back to the filing it came from. Then a comp is sound the moment you build it - not after a week of reconciling everyone's private adjustments. That is what PitchFit is for: every statement standardized one way, every number traced to source.
