horserank model trained through 2026-06-28 · UTC 06:53:13Z · refresh 60s

← back to all races

Ballarat Synthetic R2

Global Turf Maiden Plate · 1200m · 3-Y-O, Maiden, Set Weights · Synthetic · jumps · market 1.259538870

model confidence medium 0.42
Mostly pre-trading. 13 of 13 runners don't have a tradeable Betfair market yet (best back at the $1.01 floor or spread > 2×). Those rows are flagged with a PE badge — the model used the punters Edge price (the broker's own fair-odds estimate) as the market input instead. Bet suggestions are evaluated per-runner: a row can still fire BACK or LAY if that specific runner's live Betfair spread is tight enough — typically only the few favourites with serious money on them.
# horse jockey trainer b punters SP live back live lay live LTP model $ edge pp EV % suggest
1 Bohemian Star John Allen Vincent Malady 12 12.00 34.00 PE10.08
3 Dunnenkate Bro Jake Duffy Craig Weeding 4 15.00 580.00 PE21.36
4 Foxtrot Lad Mitchell Aitken Vincent Malady 9 9.50 32.00 PE7.63
9 Lady In Red E Walsh Shay Keating 15 31.00 580.00 PE46.46
11 Natural Event Craig Newitt Jody Thompson 2 41.00 200.00 PE50.83
2 Brilliant Bull Ant Cory Parish Olivia Osborne 14 26.00 580.00 PE46.46
5 Sibun Alana Kelly Mitchell Freedman 6 15.00 580.00 PE21.36
6 Sonic Surge Billy Egan Lloyd Kennewell 11 8.00 2.00 PE7.63
7 Apache Sunrise Cian MacRedmond Archie Alexander 3 5.00 580.00 PE4.48
10 Lovely Head Jye McNeil Mitchell Leek 8 51.00 PE74.00
12 Queens Of Vino Teo Nugent Ben, Will & Jd Hayes 13 9.50 7.73 PE10.08
14 Cairo Princess Sarah Robbins Jody Thompson 1 34.00 150.00 PE46.46
15 Kiss The King Jye McNeil Ben, Will & Jd Hayes 7 9.00 580.00 PE8.02

column glossary

confidence
The badge next to the race title (high medium low) blends three signals: (a) sharpness — how peaked the model's predictions are (a flat 1/N across runners = no information = 0); (b) price quality — fraction of runners with a real live Betfair market (PE-fallback counts at 0.5); (c) form coverage — fraction of runners with a prior race on record. Hover the badge to see the breakdown. Treat low-confidence races as informational — the model is making weak claims.
#
runner number assigned by punters.
horse
name (links out to punters.com.au).
jockey / trainer
connections; if not yet known.
b
barrier (gate) number.
punters SP
final starting price from punters. for pre-race runners; populated after the race finishes.
live back
best price available to BACK on Betfair right now. If you click BACK at this price, the bet is matched immediately.
live lay
best price available to LAY on Betfair right now. To lay, you accept this price (and the liability that comes with it).
live LTP
last traded price — the price of the most recent matched bet on Betfair. if nothing has traded yet.
model $
model's fair decimal-odds estimate — the price at which the bet would be break-even before commission. Calibrated against realised wins via the v2 LightGBM ensemble + isotonic recalibration, summed-to-1 across the race. To trade profitably you need a live back HIGHER than this (for BACK) or a live lay LOWER than this (for LAY), with margin to cover commission. A PE badge means the live Betfair market wasn't tradeable yet so the model used the punters Edge price (the broker's fair-odds estimate) as input. A ? badge means we had no market price at all — predictions are uninformed.
edge pp
percentage points above the commission-adjusted break-even threshold at the price we'd actually trade at (live back for BACK, live lay for LAY). Must clear +2pp before a side is suggested.
EV %
expected return on $1 staked, assuming the model's win probability is the true probability. After 5% Betfair AU commission on net wins.
suggest
BACK = back at the live back price (model thinks horse is MORE likely than the price implies, with enough edge after commission). LAY = lay at the live lay price (model thinks horse is LESS likely than the price implies). = no edge above the threshold; stand aside. Wide spreads usually produce STAND because both sides cost too much.

Not financial advice. Bet sizing isn't shown — fractional Kelly at ≤10% of full Kelly is sane, with a per-bet cap and liquidity check before sending anything. The model is recalibrated each night by the daily pipeline; it lags new horses and unusual track conditions.