Find the race.
Choose the date and meeting, then tap a race. The start time, captured field and saved-pick availability are clearly labelled.
Australian racing / Paper research
PICK A RACE.
CHECK THE
FIELD.
See the runners, get the model’s picks and compare recorded odds. Try another model without losing your first forecast.
Experimental estimates. No bets placed. No guaranteed winners.
Adults 18+. Only spend what you can afford to lose.
THE FIELD / PREDICTA
TRACK GRAPHIC. NOT RACE DATA.18+ / PAPER ONLY
Start here
From meeting list
to saved picks.
Choose a meeting and race. Request picks from the selected model. Check the results when the race has settled.
Compare the picks.
See the primary winner-model pick and top selections. Choose another approved model for a separate forecast, not a blended answer.
Check the result.
Saved forecasts keep their model and capture time. Verified results are counted later; pending or excluded races aren’t assumed losses.
Race view and day report
One race.
Or the full card.
Open a single race to inspect the field. Use the day report to read available forecasts in start order.
- 01Your race, without the clutter
The leading selection comes first. Open runner comparisons and technical detail only when you want them.
- 02A saved plan for the day
The day report groups available forecasts and labels their model, timing and result status. New reports use the model you choose.
- 03No value? No pressure.
The experimental value selector can abstain. A positive estimate is not a promise of profit, or a reason to bet.
Chance and price are different questions
Win estimate.
Recorded price.
Predicta compares an experimental win estimate with a recorded TAB win quote. That can help explain a possible angle. It cannot predict your bank balance.
A maths example, not a selection
What does estimated value mean?
If a model estimated a 25% win chance and the recorded decimal quote were 5.00, the calculation would be:
0.25×5.00− 1 =+25%
A theoretical expected return per unit, not a predicted profit on this race. The estimate can be wrong; actual results and costs can differ.
Verified saved results
What happened
to the saved picks?
See how saved pre-race selections performed against verified, matched TAB results. Training scores are not counted as race-day results.
Loading verified performance. No percentage is shown until eligible results are available.
How these figures are counted
One earliest saved forecast per model and race. Winner strike rate is correct unique winner picks divided by eligible settled races. Tied top picks do not enter that denominator. Late scratchings, ambiguous results, invalid timing and failed source checks are excluded. Pending results are not losses.
Exact first-three order is a separate finishing-order output, not the primary winner selections or a paid-place bet. Historical models have different race samples and may overlap, so these percentages are not a head-to-head comparison. Retrained model versions start their own record.
Model versions are experimental and their training and saved-forecast samples differ. Observed strike rate is not proof of future accuracy, calibrated probabilities or a profitable betting edge. This section reads saved evidence only and refreshes approximately every five minutes after result matching.
Good to know
What to know.
Models can be wrong. These are the limits.
Will Predicta tell me the winner?
It gives you the model’s leading selections, not certainty. Horse racing is unpredictable. A top-ranked runner can lose, and model rankings are not an exact first, second and third finishing order.
What goes into the predictions?
The currently selected model uses historical horse, jockey, trainer and pairing observations. Predicta also displays available TAB race-card information and odds. Inputs vary by model: displaying a field does not mean every displayed item was used in training. The application shows the selected model and its data limitations.
Are top-three chances the same as a place bet?
No. A fixed top-three shortlist is separate from TAB tote paid places, which depend on the market’s declared place terms. Do not treat a top-three selection as a prediction that a particular place bet will pay.
Does Predicta place bets or guarantee profits?
No. Predicta is a race-insights and research application. It does not place bets, manage a betting account or guarantee a return. Paper ROI records are hypothetical and are not real betting receipts.
When can I generate a forecast?
The current capture window is 8 am to 9 pm Brisbane time. Allow more than ten minutes before the stated race start, and enough time for source retrieval. Forecasts can be unavailable if timing, source quality, matching or model compatibility checks fail. A day report requested before 8 am waits for the window to open.
Is Predicta affiliated with TAB?
No affiliation, endorsement or partnership with TAB is claimed. TAB is the source of race information displayed by the application. Predicta is a separate application for personal racing research.
Ready to look at a field?
Choose a meeting.
Open a race.
The picks stay saved with their model and capture time.
Keep racing a pastime, not a plan for income.
Set a time and spending limit before you start. Never chase losses, borrow to gamble or spend money you need for essentials. It is always fine to skip a race or stop for the day.
Need support? Gambling Help Online · 1800 858 858 · BetStop self-exclusion.
Adults only. Research insights, not betting advice. No proven betting edge. Never a guaranteed winner.