Model Leaderboard
Prediction accuracy across professional tier matches · 3454/5269 settled
#1
73.6%
Vanguard
META: fires only when Elo and WR10 strongly agree (>60% same side).
89 ✓ / 121
32 wrong
#2
73.0%
Oracle
Conservative: Elo HQ dampened when streaks disagree.
73 ✓ / 100
27 wrong
#3
72.1%
Phantom
High-conviction: only commits when Elo + WR10 strongly agree.
111 ✓ / 154
43 wrong
#4
70.8%
Consensus
META: triggers when ≥75% of base models converge on same side.
211 ✓ / 298
87 wrong
#5
70.5%
Nexus
META: 3-way consensus of Apex + Oracle + Eagle with avg confidence >55%.
167 ✓ / 237
70 wrong
#6
70.0%
Roster Change
Penalty for unstable rosters — fresh lineups underperform Elo.
187 ✓ / 267
80 wrong
#7
69.5%
Elo H2H
Standard head-to-head Elo, scale=400.
228 ✓ / 328
100 wrong
#8
69.5%
Patch-Aware
Elo with patch-boundary decay — stale form loses weight after Valve patches.
221 ✓ / 318
97 wrong
#9
68.7%
Streak
Win/loss streak signal blended with Elo.
299 ✓ / 435
136 wrong
#10
68.0%
Fatigue
Dota-specific: penalty for back-to-back matches (<12h apart).
223 ✓ / 328
105 wrong
#11
65.9%
Apex
Blend: Elo + form + streak — momentum-aware ensemble pick.
207 ✓ / 314
107 wrong
#12
65.8%
Eagle
Blend: Elo + WR10 + form — weighted toward top-tier record.
208 ✓ / 316
108 wrong
#13
64.2%
SOS
Strength of schedule: form weighted by opponent quality.
314 ✓ / 489
175 wrong
#14
62.6%
WR10
Win-rate vs top-10 opponents, blended with Elo.
124 ✓ / 198
74 wrong
#15
62.2%
Form
Pure win-rate over last 10 matches.
235 ✓ / 378
143 wrong
#16
59.4%
Meta Strength
Team's recent hero pool weighted by current meta winrates.
76 ✓ / 128
52 wrong
#17
56.2%
Tier Elo
Elo computed only from matches against top-30 teams.
178 ✓ / 317
139 wrong
#18
55.2%
Synergy
Hero co-occurrence winrates within recent team lineups.
290 ✓ / 525
235 wrong
◉
Calibrating — fewer than 20 settled predictions on this tier. Ranking appears once more matches complete.
72.2%
Elo HQ
High-confidence Elo: smaller K-factor, scale=2000 — slow learner.
calibrating (18/20)
Accuracy is measured per tournament tier — Premium is the clean, high-signal subset (mirrors cs2predict's top tier). Models with fewer than 20 settled predictions on the selected tier are calibrating. Coin-flip predictions (48–52%) are excluded. Click a model for confidence buckets, Brier score, and recent picks.