Silver?Draft with context
Our model analyzes your history, team synergy, and enemy matchups to recommend picks with high predicted win rate for your situation.
YOUR TEAM
Your History
- Champion winrate
- Tag synergy
Team Synergy
- Ally pairwise WR
- Team damage balance
Enemy Matchups
- Counter stats
- Lane matchup
Champion Stats
- Base winrate
- Difficulty curve
Recommended Picks
ENEMY TEAM
Adaptive builds, every game
Our model evaluates game state (enemy comp, gold differential, team needs) and suggests an item path for your situation.
Build Path Explorer
ADC build tree with recommendations
Game State Inputs
- Enemy team composition
- Gold differential @ 15 min
- Your champion's scaling
- Team damage profile
- Objective timers
Contextual Branching
The model branches based on game context. Different games can produce different build paths.
Specific to the game
Computed from 298 features: your champion and items, gold/XP, objectives, lane state, and team comps. As the game changes, the suggestions can change too.
Cause, not correlation
We use causal inference (DoubleML) to estimate how different behaviors affect rank.
Average Treatment Effects
Causal effects estimated from large-scale match data
CATE vs ATE
Conditional treatment effects
ATE tells you what works on average. CATE tells you what works for you. Using DoubleML, we estimate individual-level causal effects conditioned on your playstyle, champion pool, and rank.
37 Behavioral Signals
Champion pool, builds, timing, tilt
We compute a set of behavioral signals from your match history: champion focus, build choices, skill order patterns, play-after-loss behavior, and more.
Correlation Misleads
Correlation mixes skill and cause
Correlations often say "good players do this," not "doing this makes you win." Causal estimates aim for the second, controlling for skill and context.
Built for power users
Fast queries across hundreds of millions of matches. Full SQL access for custom analysis. Pivot and chart results in the console.
- Query Console with autocomplete
- Custom views and charts
- Real-time Riot API integration
- Multiple regions, large datasets
SELECT champion_name,
AVG(kills) AS avg_kills,
AVG(deaths) AS avg_deaths,
COUNT(*) AS games
FROM summoner_match
WHERE puuid = 'your_puuid'
GROUP BY champion_name
ORDER BY games DESC
LIMIT 10;Get the desktop app
A standalone Windows app. Not an Overwolf app.
Free to use. No account required to get started.