EXHIBIT A
Matchup Draw
A full-deck win-chance model scores every battle, then compares your draw against the live meta panel — the decks a fair matchmaker would actually hand you, weighted by real play frequency.
AnalysisAnalyze
Analysis request
Enter your player tag. Receive the evidence. Build your case*.
Pull an unsuspecting player from the files at random.
*Admissible in no court. Devastating in clan chat.
CLASSIFIED - CASE FILE #775
LIVE FROM THE CASE DESK
775 files on record
| Player | Luck |
|---|---|
| nerf minerSuspiciously Average | 57 |
| ArmoShawMain Character Draws | 84 |
| CODE SweatyCoin-Flip Normal | 52 |
| ilhvnxLow-Key Lucky | 64 |
| poraxSuspiciously Average | 53 |
| MiloCoin-Flip Normal | 50 |
EXHIBIT A
A full-deck win-chance model scores every battle, then compares your draw against the live meta panel — the decks a fair matchmaker would actually hand you, weighted by real play frequency.
EXHIBIT B
Buckets every game by the win streak you were riding and flags fresh deck changes. If win chances sink exactly when you climb, the timing itself is the evidence.
EXHIBIT C
Normalized level gaps remain supporting evidence. If levels are even, this factor automatically loses influence.
EXHIBIT D
A separate skill score measures results against the model's predicted win chances: 50 means you played to the cards, higher means the hands are doing work.
A tag report is the way in. These are the rest of the tools built on the same model.
Build a deck and score all 8 cards together against the live meta, then let the model try swaps. It shows only changes that improve the complete deck.
The decks the model rates highest right now, for ranked and for Trophy Road, with their measured win rates.
Per-card usage, win rate, counters and synergies — and the decks currently carrying each card.
Every report on file, the luckiest and unluckiest of the day, and what other players make of them.
The correlations behind the model: what actually moves a win rate, and what only looks like it does.
The same model behind every page, as REST endpoints: deck scoring, matchups, meta panels. Interactive docs and keys live in your account.
Support the investigation
Higher daily limits, upgrade searches limited to your collection, score history, and the upgrade planner.
Each report analyzes up to the newest 200 usable 1v1 battles, factors weighted by evidence and pulled toward neutral when it's thin. Modes (ranked, ladder, Clan War, events) can be isolated; team battles excluded.