tarai+

Statistical rankings that earn their keep on race day.

Four modern models score every runner. Race history fuels evaluation. A final report puts prediction against the result.

View the models

Four independent views of the same field.

Each model ranks runners with a different statistical lens — then you compare them side by side.

Weighted scoring

Composite form factors blended into a transparent weighted score and rank.

Logistic regression

Probabilistic win hypothesis from calibrated LogReg estimates.

LightGBM

Gradient-boosted trees capturing non-linear patterns in race history.

RRF fusion

Reciprocal Rank Fusion merges model signals into a consensus order.

History in. Rankings out. Results measured.

The workflow is built for continuous evaluation — not a one-off tip sheet.

Collect race history

Ingest meetings, entries, and past runs so every model sees the same evidence.

Produce four rankings

Weighted, LogReg, LightGBM, and RRF each publish an ordered win hypothesis.

Record the result

Official finishing order is captured alongside the pre-race rankings.

Evaluate & report

Hit rates and model agreement surface in the race ranking report.

Actual finish versus every ranking.

Export a final report that lines up the race result with Weighted, LogReg, LightGBM, and RRF — so you can see which hypothesis held.

Actual result
Official
Weighted
Align
LogReg
Align
LightGBM
Align
RRF
Align

Ready to rank the next meeting?

Open race cards, run the four models, and export the comparison when the result is in.

An unhandled error has occurred. Reload 🗙