Probability is not certainty
A 25% chance should lose roughly three times out of four. One result cannot validate or disprove an estimate.
Educational simulator · Race 7 · 1 mile dirt
Read the race. Price the chance. Review the decision.
Build a virtual wager, see the probability and expected value before you commit, then study what the result does—and does not—say about your process.
- Bankroll
- 1,000.00 cr
- Race risk cap
- 20.00 cr
- Races studied
- 0
- Versions
- — · —
Choose your exercise
How much help do you want before the race?
01
Before the race
Turn the racecard into a probability
Every rating is on a 0–100 scale. The lab combines five visible factors, then converts each total into a share of the field’s win probability. No horse is guaranteed and your bet never changes the result.
32%Speed
25%Recent form
18%Class
15%Pace fit
10%Post draw
The weighted scores feed a Plackett–Luce ranking model. After a horse is selected for one finishing position, the remaining probabilities are renormalized for the next. The displayed place, show, and exotic probabilities are exact sums across all 720 possible finishing orders.
Fairview Park · Race 7
Allowance · 1 mile · Fast dirt
Balanced field
80+ 70–79 under 70
| Runner | Style | Speed | Form | Class | Pace fit | Draw | Model win | Win odds |
|---|
Model win is this lab’s estimate. Win odds are the posted virtual price. A difference between the two creates positive or negative expected value.
Simulation
The stretch
Build at least one wager to run the race.
Post-race review
Judge the decision separately from the result
| Wager | Your estimate | Model chance | Price | Model EV / cr | Result | Net |
|---|
Keep the lesson
Four habits worth taking away
Price changes the bet
The likely winner can still be poor value. Compare your chance with the price’s implied chance before committing.
Small stakes preserve choices
A 2% race cap keeps a normal losing sequence from ending the learning session too quickly.
Review the process
Record the probability, price, and reason before the race. Grade the reasoning after it—not just the payout.
Audit trail
Versioned session record
Up to 20 device-local results. Versions and seeds keep later comparisons honest.
No completed races in this local session yet.
| Completed | Racecard | Versions | Seeds | Stake | Net |
|---|