Model Leaderboard
Prediction accuracy across premium tier matches · 1545/2218 settled
#1
80.2%
Consensus
META: triggers when ≥75% of base models converge on same side.
73 ✓ / 91
18 wrong
#2
77.4%
Apex
Blend: Elo + form + streak — momentum-aware ensemble pick.
89 ✓ / 115
26 wrong
#3
75.8%
Nexus
META: 3-way consensus of Apex + Oracle + Eagle with avg confidence >55%.
72 ✓ / 95
23 wrong
#4
74.7%
Streak
Win/loss streak signal blended with Elo.
127 ✓ / 170
43 wrong
#5
74.5%
Oracle
Conservative: Elo HQ dampened when streaks disagree.
38 ✓ / 51
13 wrong
#6
72.1%
Vanguard
META: fires only when Elo and WR10 strongly agree (>60% same side).
62 ✓ / 86
24 wrong
#7
71.2%
Eagle
Blend: Elo + WR10 + form — weighted toward top-tier record.
94 ✓ / 132
38 wrong
#8
71.0%
Elo H2H
Standard head-to-head Elo, scale=400.
98 ✓ / 138
40 wrong
#9
70.9%
SOS
Strength of schedule: form weighted by opponent quality.
100 ✓ / 141
41 wrong
#10
70.9%
Fatigue
Dota-specific: penalty for back-to-back matches (<12h apart).
100 ✓ / 141
41 wrong
#11
70.9%
Roster Change
Penalty for unstable rosters — fresh lineups underperform Elo.
90 ✓ / 127
37 wrong
#12
69.9%
Form
Pure win-rate over last 10 matches.
86 ✓ / 123
37 wrong
#13
69.4%
Phantom
High-conviction: only commits when Elo + WR10 strongly agree.
77 ✓ / 111
34 wrong
#14
69.4%
Patch-Aware
Elo with patch-boundary decay — stale form loses weight after Valve patches.
93 ✓ / 134
41 wrong
#15
65.5%
WR10
Win-rate vs top-10 opponents, blended with Elo.
112 ✓ / 171
59 wrong
#16
62.2%
Tier Elo
Elo computed only from matches against top-30 teams.
89 ✓ / 143
54 wrong
#17
61.5%
Meta Strength
Team's recent hero pool weighted by current meta winrates.
16 ✓ / 26
10 wrong
#18
56.2%
Synergy
Hero co-occurrence winrates within recent team lineups.
118 ✓ / 210
92 wrong
◉
Calibrating — fewer than 20 settled predictions on this tier. Ranking appears once more matches complete.
84.6%
Elo HQ
High-confidence Elo: smaller K-factor, scale=2000 — slow learner.
calibrating (13/20)
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.