Selection confidence and betting value are separate. DOS scores form, place reliability, distance, class, weight, barrier/map and track/going only when that evidence exists. It then calculates a conservative paper-test place probability, fair odds and expected value. A high-confidence runner is still a PASS when the market price is too short.
| Jump | Meeting | Race | Name | Distance | Mode |
|---|
Review the whole Australian thoroughbred card at any time. This is an investigation view only — it does not place bets or interfere with Auto Mode.
| Time | Runner | Market | Stake | Price | Confidence | Fair | Edge/EV | Status |
|---|---|---|---|---|---|---|---|---|
| No automatic paper bets yet. | ||||||||
Use this at the track or from home. Select today's race and run the latest analysis. It can recommend WIN, PLACE, exotics or PASS. Suggestions are advice only; they do not bypass the three live-execution checkpoints.
For Flemington Derby Day and Melbourne Cup Day the app switches from DOS Classic to the broader Full AI model. The default priority is WIN and PLACE. Quinellas, exactas and trifectas are treated as selective research bets only unless the feed contains enough live market/pool information to demonstrate value.
WIN • PLACE • WIN/PLACE
Quinella • Exacta • Trifecta — only when justified
Up to $100 per race; PASS remains valid
Do not infer tote pool value when pool/dividend data is absent.
Start by measuring WIN and PLACE performance cleanly. Record exotic ideas separately so the Research Lab can compare their theoretical results without allowing unsupported exotic bets into automatic execution.
Start with $2.00. Five Time is now a separate market/form experiment — not DOS Classic. It looks for suitable non-maiden races and aims around $2.00 PLACE odds. If a leg places, the full return becomes the next stake; a loss stops the run.
$1.70–$2.40 • target $2.00
8–12 runners • maiden races excluded
No minimum DOS score • no coverage minimum • no EV gate
Market first • form-guide signals where supplied • late information/news to be measured separately, not blindly trusted
| Leg | Jump | Meeting/Race | Selection | PLACE | Model | Stake | Result | Return |
|---|---|---|---|---|---|---|---|---|
| Choose a date to test the system. | ||||||||
The code path is prepared for live use, but Five Time is deliberately not armed by default. Before live activation, Research Lab should validate the fixed rules on a meaningful out-of-sample history. The live runner must also re-check price, scratching, daily limits and the rolling stake before every leg.
Check what the configured racing API can actually supply before running a backtest. This separates working feeds from plan-blocked historical prices.
Fetch official settled results for a chosen date from the configured racing provider. This imports finishing positions and settles matching existing paper bets. It does not invent pre-race DOS selections or historical odds. Original recorded Five Time requires pre-race snapshots; historical simulations require accessible archived odds.
The research agent continuously captures races, searches for repeatable patterns, benchmarks DOS-style rules against modern models, paper-tests candidates and freezes only strategies that pass validation. The final 12-month holdout remains untouched during discovery.
DOS WIN, DOS PLACE and original Five Time use only genuine recorded pre-race analyses. If none were captured, the tester reports NO RECORDED PRE-RACE DATA rather than zero qualifiers. Historical Review requests archived pre-jump PLACE prices through the configured API. PuntersEdge Free/Hobby keys do not have closing-line archive access; Standard has a rolling 90-day window. Results alone cannot reproduce historical odds. It does NOT reproduce DOS scoring or actual past tips. Up to 7 days per archive run; other tests up to 90 days.
Compare staking behaviour on the same five selections. This does not change live staking.
Roll the full return through all five legs.
After leg 3, bank the original $2 and continue with the balance.
After leg 3, bank 50% and continue with 50%.
After a completed 5/5 run, test the next sequence starting at $4.
Use Historical Race Review/Strategy Tester first; the lab compares staking only after genuine selections exist.
Ask the Research Agent in normal English. It can understand ranges such as “last two weeks”, test Five Time 2.0 historical candidates, compare rule changes, and reply with the horses and results. Historical questions use the racing API configured on the server automatically.
| Date | System | Races/legs | Status | Start | Return | P/L |
|---|---|---|---|---|---|---|
| Run a test to see results. | ||||||
Shows the recommendation stored before the race — including PASS decisions — whether or not you followed it. Results and theoretical payouts are added when the result feed has settled the race.
Separate from genuine Emuvale History. Fetch a completed day’s archived market/results data and simulate Five Time 2.0. Retrospective results are always labelled as simulations.
| Jump | Race | Agent | Suggestion | Stake @ price | Conf. | DOS value | Result | Return | P/L | Paper? |
|---|
When you're at a meeting, use Trackside/LIVE TODAY to see the current suggestions. The Bank page shows the limits that the Risk Agent will enforce automatically.