Live chase win-probability.

Two ways in, pick yours.

The signal, in full.

Pass the match state; receive a complete, model-backed read.

1. 01

   Calibrated win probability

   A true, calibrated probability the chasing side wins from the exact match state, trained with log-loss, validated on Brier score. Beats the run-rate heuristic by 8.6 percentage points.

   chasing_team_win_probdefending_team_win_prob

2. 02

   Full situation read

   Runs needed, balls remaining, current and required run rates, and wickets in hand, the complete state returned alongside the probability.

   runs_neededballs_remainingrequired_rr

3. 03

   Phase-aware confidence

   Appropriately uncertain in the powerplay and confident by the death overs, accuracy climbs from 73% to 88% as the chase resolves.

   current_rrwickets_in_hand

Learned from 475,000 live ball-states.

Fit on a large archive of completed professional matches, then validated on unseen games. The intelligence lives in the model.

• 475,000 ball-states

  Every ball of 4,406 real completed chases, labelled with the eventual result.

  trained

• Gradient-boosted model

  XGBoost over match-state features, not a run-rate rule of thumb.

  proprietary methodology

• Calibrated, not just ranked

  Trained with log-loss and validated on Brier score, so a stated 70% genuinely means ~70%.

  calibrated

• Out-of-sample AUC 0.88

  Measured on matches held out of training.

  out-of-sample

What comes back.

{
  "chasing_team_win_prob": 0.489,
  "defending_team_win_prob": 0.511,
  "situation": {
    "runs_needed": 60,
    "balls_remaining": 30,
    "current_rr": 6.67,
    "required_rr": 12,
    "wickets_in_hand": 8
  }
}

Use cases.