Chess Opening Statistics

Explore opening families ranked by expected score. Filter by side to see what you can play as White or Black.

Source Lichess·Games 2.1B·Range 2023-12 to 2025-11·Updated January 2026

Games Analyzed0.0B
Opening Families0
Rating Brackets00 to 2500+
Time Controls0UltraBullet to Classical
Why This Matters

Openings are side-specific. So are the expected scores.

Results differ by side and by rating bracket—use the filters below to see what actually wins in your games.

Browse Openings

56 families ranked by White expected score

Filter Options
You Play
Rating
Time
Sort

These are opening families. Some variations require opponent cooperation — click into any opening to see the full breakdown.

Total56
Games1.0B
Expected Score50.8%
Avg Edge+1.3%

56 families

Page 1 of 3

What are the best chess openings for White and Black?

Snapshot

TrueElo analyzed 2.1B rated Lichess games (2023-12 to 2025-11) across 131 opening families. Rank openings from your perspective (playing White or playing Black), then break it down by rating bracket and time control. Every win/draw/loss estimate includes a 90% Bayesian credible interval and sample size.

Methodology
Data Source
2.1B+ rated games from Lichess (2023-12 to 2025-11). Game data is owned by Lichess (CC0 license). TrueElo provides the statistical analysis only.
Filtering Criteria
  • White-centric: bucketed by White’s rating bracket
  • Opponent within 200 Elo (rating gap filter)
  • Minimum 2 moves played
  • Rated games only (no casual)
Statistical Method
Bayesian inference with a 3-outcome model (white win / draw / black win). We report 90% credible intervals. More games = tighter intervals = more reliable estimates.
Minimum Sample
Opening pages: n ≥ 100. Gambit breakpoints: n ≥ 50. Complexity/traps: n ≥ 500.

Frequently Asked Questions

What data is this based on?

All statistics are derived from 2.1B rated games from Lichess, covering 131 unique openings across 9 rating brackets and 6 time controls.

How are confidence intervals calculated?

We use Bayesian statistics with a 3-outcome model (white win / draw / black win). Each displayed 90% confidence interval comes from the marginal distribution of that outcome—more games = narrower intervals = more reliable estimates.

Why do win rates differ from other sources?

Our analysis filters for quality: games are bucketed by White’s rating bracket, filtered to opponents within 200 Elo, and include all games with at least 2 moves. This reduces noise from mismatches while including short decisive games like early checkmates and resignations.

How often is this data updated?

Statistics are regenerated periodically from the latest Lichess public database. Check the "Last Updated" date on each page for the data freshness.