Pairs tier list

Tiers rank raw observed win percentages within this category, not across categories. S: top 10%; A: next 20%; B: middle 40%; C: next 20%; D: bottom 10%.

Equal raw rates stay together at their midpoint percentile. Some tiers can be empty. The displayed rate rounds to one decimal place.

Entries need at least 100 observations. A category needs at least 5 eligible entries. Other entries stay unranked.

The 100-observation minimum is a display rule, not a statistical reliability threshold.

Search and the tier filter hide entries without new tier assignments. Tiers describe observed performance, not popularity or predicted results.

Window: 2026-09-24 00:00:00 UTC to 2026-10-23 00:00:00 UTC.

The date describes the latest observation for this kind. Other observations in the window can be older.

The sample covers 30 calendar days, ranked mode, and Balance 26. Patch attribution is inferred from public profiles.

An observation is a round record with this entity or pair. Records include reconstructed terminal boards from public history.

Public profiles do not cover every player. Rarity, round, lineup, player skill, and selection can affect observed rates.

The win percentage divides wins by observations. It describes these matches, not the effect of the entity on a result.

The same run can contribute repeated observations. Do not add row counts to estimate unique players, boards, or runs.

The sample includes at most 100 pairs, selected by observation count. It is not a complete list or proof of causal synergy.

Other windows, modes, rank bands, and round bands are not available. Private Companion matches are not part of this sample.