CritRate · player performance metrics
Player performance metrics, computed from the match as it runs.
Status: In development
CritRate is a 1 to 10 rating for every player, every match, computed from in-game performance as it happens. No manual scoring, no panel of pundits. It is built on the same detection and event pipeline as CritShot, which is live on broadcast today, so the readings it rates are ones the pipeline already produces on every ball.
CritRate · In development
1 to 10. Per player, per match.
Computed from in-game performance data as the match runs, with no manual input. Built on the same detection and event pipeline as CritShot, trained on the same 30 TB dataset from PSL, National T20 and Tri-Nation cricket.
Each bar is one rating band; the marked band is where the current innings sits.
Inputs
The readings a rating is built from already exist.
CritRate does not add a measurement pass to the fixture. Every reading below is produced by the CritShot pipeline on each delivery, on the same workstation at the ground, and the rating is being built on top of them.
- Who and where, on every frame
- Ball, batter, bowler, fielders and pitch are located on every frame by three detectors voting in parallel.
- What the delivery did
- Spin, speed, swing, seam and more than a dozen other features extracted from the delivery.
- How hard the ball was to play
- An LSTM and XGBoost stack picks the highest-value shot for that delivery and scores its difficulty from 1 to 10.
- The state of the match around it
- Over, phase of the innings, required rate and intent are part of the model input, alongside fielder placement and density mapped live.
Which of these readings carry the most weight in the finished number, and in what proportion, is a modelling decision that is not settled. It will be published with the metric rather than described in advance, for the same reason the detection figure is published with its method: a weighting nobody can check is not worth quoting.
Status
What is built, and what is still being built.
CritRate is In development. The parts of it that run in production today run there as part of CritShot, which is live.
| Component | Status |
|---|---|
| Ball and player detection | Built. Running in production inside CritShot, which is live. |
| Per-delivery feature extraction | Built. Spin, speed, swing, seam and more than a dozen other features per ball. |
| Shot difficulty scoring | Built. Every delivery is scored from 1 to 10 for difficulty inside the CritShot prediction stack. |
| Player rating model | In development. Turning the per-delivery readings into one number per player is the open work. |
| The app shown on this page | Concept. A pre-MVP render, with fictional teams and players. |
| Use on a live broadcast | Not yet. CritRate has not been carried on a live broadcast. |
The last two rows are the ones worth reading twice. A rating that goes on air changes how a player is talked about, so the number has to be defensible before it is broadcast, and nothing here describes it as finished.
Questions
Before you get in touch.
Can we use CritRate this season?
No. CritRate is In development. Detection, tracking and per-delivery feature extraction are built and in production inside CritShot, which is live; the rating model and the app around it are not finished. What is open today is the pilot list.
What does the number describe?
One player, in one match, on a 1 to 10 scale, computed from in-game performance as the match runs. It is a per-match reading rather than a career index or a ranking across competitions.
What would it need from our ground?
CritRate is built on the same detection and event pipeline as CritShot, so what it inherits is that pipeline's requirement: a clean feed it can read at the ground. A CritRate-specific deployment specification is settled once the rating model is finished.
What does the pilot list commit us to?
Nothing. It is a list of competitions and broadcasters who want to see the metric when there is something to see, and who are willing to say what it would have to answer to be useful to them.
The pipeline underneath
CritRate, In development, reads what CritShot already produces.
CritRate is built on the same detection and event pipeline as CritShot, which is live: the same detectors, the same per-delivery readings, the same workstation at the ground. That is the reason an in-development metric is worth a conversation now, and it is also the fastest way to judge whether the rating will hold up. Read what the pipeline does on every ball, then decide.