Won 207 of 408 matches in this window, holding tier E. Weighing the tiers on both sides of each of those matches, the model expected about 229 wins - so subak is -21.8 against what the tiers predicted.
subak wins fewer than tier E predicts, so the tier is too high. Luck alone explains a gap this big less than 1 time in 20.
CONSIDER MOVING DOWN: E → A
| Going in | Matches | Won | Lost |
|---|---|---|---|
| Favoured | 228 | 138 | 90 |
| Underdog | 104 | 34 | 70 |
| Evenly matched | 76 | 35 | 41 |
Favoured means the model gave their team better than an even chance before the match. A record built on favoured games is not the same as one built against the odds, and the totals hide the difference.
Covering 2025-09-19 to 2026-09-18, 4580 matches.
Built 2026-09-19T07:59:07+00:00. Model fitted 2026-09-19 on 6394 matches of 3v3 history, 2024-12-04 to 2026-09-19.