How the diagnosis works

OW Comp Lab lets you enter both teams’ Overwatch 2 compositions and visualizes each side’s characteristics, strengths and weaknesses, along with how the two match up. Its purpose is to show where the reasons for winning or losing may arise — not to predict win rates or to coach your play.

It never outputs a win rate, a total score or an overall rating. What you see is a set of independent traits, deliberately never combined into a single number.

Each trait is shown on a 1–5 scale. It is a characteristic, not a grade — a low band means a trade-off, not a weakness.

Traits have a polarity. Most are “higher is better”; those marked ⚠ are risks where higher is worse; some are neutral, with no better or worse.

There are two kinds of calculation. Continuous traits are graded by where the team sits among every legal composition — a percentile position. Discrete traits are graded by fixed rules based on counts or thresholds.

Formations are captured as four styles (Dive, Brawl, Poke, Rush). Each hero has an aptitude for each style; the team totals are standardized as z-scores so that styles measured on different baselines can be compared. The highest is the primary tendency.

When the top two are within 0.5σ of each other, both are shown as a dual tendency — a neutral read meaning a flexible, convertible comp, not indecision.

After the tendency is set, the tool checks for missing pieces — structural gaps that keep the comp from executing — and softer aptitude notes.

Each hero carries threat tags and weakness tags. The tool matches every enemy threat against your weaknesses (unfavorable) and every one of your threats against enemy weaknesses (favorable), each rated 1–3 for severity.

Matchups are computed at the finest grain — each (rule, source hero, target hero) counted separately — then merged into one named line per rule for display.

Some effects adjust severity: a Kiriko cleanse reduces anti-heal, hard-CC and ability-lockout matchups against her team by one.

When both teams are full, the tool tries replacing each of your five heroes with every same-role hero not already on your team, keeping the 1 tank / 2 damage / 2 support structure.

For each candidate it computes an improvement score, defined exactly as:

All differences are measured at the finest grain. Comp weaknesses are weighted double because they are structural.

Only swaps with a positive score are listed; the top three are shown (all ties included). Each suggestion states the threats it removes, the edges it gains and what it costs in exchange.

The score itself is never shown — it is only an internal ranking. The interface presents the reasons and trade-offs, never a number, in keeping with the principle of not handing you a black-box figure.

The diagnosis is based on the following data.

Data version: v33 · compositions used for calibration: 351,624 · calibrated on 2026-07-06

The underlying hero data is maintained by the site owner. The diagnostic parameters — percentile thresholds and formation baselines — are recalibrated across the entire composition space whenever the data changes.

This is an unofficial, fan-made tool and is not affiliated with Blizzard Entertainment. Overwatch and related trademarks are the property of Blizzard Entertainment, Inc.