How to choose a hotel's CompSet (competitive set)

The principle: comparable is not "nearby" or "looks similar"

The CompSet (competitive set) is the group of hotels a given asset's performance and value are measured against. It is the single piece that distorts a manual valuation the most: picking peers by intuition — "nearby hotels" or "hotels that sound similar" — is the most common mistake when someone values a hotel with a general-purpose AI that was never given the sector's rules.

A correct CompSet requires matching several axes at once — no single axis is enough to declare comparability.

1 · Class and category

An economy hotel (2-3 star) and a luxury one (5 star) do not share a revenue or cost structure even on the same street: ADR, F&B mix and GOP% are of a different nature. Class is the first filter, not just one of many.

2 · Submarket, not just city

"Madrid" is not a submarket: the historic centre, the business district and the periphery carry different ADRs and cap rates. A peer in the same city but several kilometres away, in an area with different demand, is not comparable even if it shares the class — the CompSet needs to be scoped to the real submarket, not the city label.

3 · Size and asset condition

Size (room count) drives the fixed-cost structure and which operator type can run the asset. Condition — recently renovated, in normal operation, or in need of CAPEX — moves achievable ADR and the cap rate the market demands: an un-renovated hotel on the same street as a freshly renovated one is not the same product in the market's eyes.

4 · CompSet average vs. the subject: the tell of a broken CompSet

A simple check to detect a badly built CompSet: the CompSet's average ADR should sit in a reasonable band around the subject hotel's ADR (not far above or below), unless there is an explicit reason (the subject needs repositioning, for instance). If the CompSet average sits far from the subject with no declared reason, the CompSet is mis-built — the asset is not mis-valued.

A real, dated example: Spain's prime hotel yield held at 5.0% in Madrid and Barcelona through 2025 and early 2026 (source: CBRE, "Figures Hoteles", Q3 2025 and Q1 2026). If an upscale asset in central Madrid is benchmarked against a CompSet of economy hotels on the periphery, the resulting cap rate and ADR reference say nothing useful about the real asset — the error is not in the valuation method, it is in the input CompSet.

How HotelVALORA solves this

The engine builds the CompSet algorithmically, balancing class, distance and ADR at once — not by geographic proximity alone — and shows it as an editable part of the report. The report also discloses how many real peers the asset has (which can be low in an illiquid market) instead of padding the table with poorly relevant comparables just to fill it.