Company
About Racetrackstar
Racetrackstar is a research operation for US thoroughbred racing. Its staff builds each card from entries, prices it against the tote, and grades it against the official chart.
The Team
Racetrackstar is run by handicappers and quantitative analysts, with a data engineering group that keeps every input current through post time. The handicappers frame the questions a race poses. The analysts decide whether the evidence answers them, and the price decides whether the answer is worth a wager.
Each card publishes before the race and is not edited afterward.
What Goes Into a Race
Public data on a starter is spread across charts, entries, workout tabs, and the tote. Racetrackstar joins it into one record per runner before the field is ranked.
| Input | What is read | What it changes |
|---|---|---|
| Chart history | Early position, late gain, trip trouble, and wide trips across recent starts | Pace shape, and a fairer read of a dull last line |
| Late changes | Jockey, equipment, medication, and weight changes posted on race day | Whether the morning read still holds at post |
| Workouts | Recency, frequency, and pattern on the work tab | Readiness off a layoff or a freshening |
| Conditions | Surface, going, off-turf moves, rail position, and weather toward post | Which form lines apply today |
| Pedigree | Sire aptitude by surface and distance | First turf starts and sprint-to-route switches |
| Market | Morning line, live tote, and closing price | Where the model and the public disagree |
When an input is missing for a runner, the model adds nothing for it. It does not estimate the gap.
The Research Lab
The research lab tests every model change before it reaches a published card. Its work runs on two ledgers that cannot be rewritten after the fact.
| Ledger | Records | Used for |
|---|---|---|
| Factor ledger | Every start, with the factor scores held at decision time | Separating factors that predict finishes from factors that only describe them |
| Play ledger | Every system play, with the price pinned when the decision was made | Return, hit rate, and closing-line value |
Win probabilities are scored against the public pool on the same races, by track and by surface. A factor change ships only when it improves that score on races it was not fitted to. When a sample is too small to judge, the lab withholds the result.
Calibration is a target, not a guarantee. A runner published at 20 percent should win about one race in five over a large sample, and no single race proves or disproves it.
Pro on Race Day
The morning card is free and frozen at publication. Pro is the race-day layer, built for the hours when fields and prices change.
| Tool | Function |
|---|---|
| Live board | Live odds, scratches, and cancellations through post |
| Race queue | Followed races across every track, in post-time order |
| Breakdown Lab | The component scores behind each runner's rating |
| Lab sliders | Reweight one race and compare your order with the published one |
| Circuit Pad | A stake plan for one track's card, from a bankroll you enter |
| Stake Assist | A stake plan beside the race you are reading |
Stake tools produce a plan. Copying an amount never places a wager. Pro pricing
Questions
How is Racetrackstar different from a tip sheet?
A tip sheet publishes opinions and rarely grades them. Racetrackstar publishes a probability for every runner, compares it with the price, and grades each play at the price held when the decision was made. No role is a lock, and losing months stay in the ledger.
What does the research lab test?
Whether a factor predicts finishes on races it was not fitted to. Each change is scored against the public pool by track and surface. Changes that only fit history do not ship, and results from small samples are withheld.
How are win probabilities calibrated?
Ratings become probabilities through a calibration fitted to recorded finishes, then checked against the tote. Runners published at 20 percent should win about one race in five over a large sample. The method and its parameters stay private.
Why grade at the decision price and not the closing price?
The closing price is not available when the play is made, so grading against it would flatter the record. The gap between the two, known as closing-line value, is tracked separately as a check on the read.
What does Pro add?
The race-day layer: live odds and scratches through post, a cross-track queue in post-time order, the Breakdown Lab with its sliders, and stake planning from a bankroll you enter. Copying a stake never places a wager.
Research, ledger, and Pro tool questions are answered in the full FAQ.