Spring Carnival 2026: what the data says
We analysed every race TrackTips rated from 1 October 2026 to 5 November 2026. One of our top 3 won 52.1% of them, and one of our top 3 won 13 of 18 Group 1s. But backing every pick blindly still lost money.
Method
We took every race TrackTips rated from 1 October 2026 to 5 November 2026: 1,986 races and 5,691 selections across Australian thoroughbred meetings, with scratchings removed. For each race our model scored every runner before the jump and published its top 3. We then compared those picks with the official results and starting prices (SP). All profit figures use level win bets at SP.
How each pick performed
| Pick | Runs | Won | Placed | Win return |
|---|---|---|---|---|
| Top pick | 1,859 | 21.7% | 47.9% | -21.6% |
| 2nd pick | 1,865 | 17.3% | 42.9% | -24.5% |
| 3rd pick | 1,854 | 15.3% | 40.6% | -15.4% |
Our top pick won more often than our 2nd and 3rd picks, so the ranking does what it should. But strike rate isn’t profit: the market often prices our top pick short, so the returns come from choosing which picks to back, not from backing them all.
Do higher ratings win more?
Strike rate of our top pick by rating band (bands with at least 30 runs):
The average winning starting price across our winning picks was $4.39, so the winners aren’t just favourites.
Group 1 races
Across the 18 Group 1 races in the period, one of our top 3 won 13 and placed in 17.
| Date | Race | Best result from our top 3 |
|---|---|---|
| Sat 3 Oct | Flemington R8 | Won: Cosmic Crusader (3rd pick, $8.00) |
| Sat 3 Oct | Royal Randwick R6 | Won: Global Goal (2nd pick, $3.30) |
| Sat 3 Oct | Royal Randwick R8 | Placed: Ceolwulf (3rd pick) |
| Sat 10 Oct | Caulfield R7 | Placed: Sheza Alibi (2nd pick) |
| Sat 10 Oct | Caulfield R8 | Won: Warwoven (3rd pick, $2.40) |
| Sat 10 Oct | Caulfield R9 | Won: Lady Shenandoah (top pick) |
| Sat 17 Oct | Caulfield R7 | Won: Saintly Dancer (3rd pick, $4.60) |
| Sat 17 Oct | Caulfield R9 | Placed: Saint George (3rd pick) |
| Sat 17 Oct | Royal Randwick R7 | Won: Tempted (3rd pick, $6.50) |
| Sat 17 Oct | Royal Randwick R9 | Won: Autumn Glow (top pick, $1.26) |
| Sat 24 Oct | Flemington R8 | Won: Cosmic Crusader (top pick, $9.50) |
| Sat 24 Oct | Royal Randwick R7 | Won: Proxima Dea (3rd pick, $3.70) |
| Sat 31 Oct | Flemington R6 | Placed: St Gotthard (top pick) |
| Sat 31 Oct | Flemington R7 | Won: Stormy Marco (top pick, $4.40) |
| Sat 31 Oct | Flemington R8 | Won: Lady Shenandoah (top pick, $1.85) |
| Sat 31 Oct | Royal Randwick R8 | Won: Skyhook (2nd pick, $11.00) |
| Tue 3 Nov | Flemington R7 | Unplaced |
| Thu 5 Nov | Flemington R8 | Won: Proxima Dea (2nd pick, $2.10) |
Backing every pick blindly
Backing every one of our top 3 picks with no filter returned -20.5% on turnover. That’s the most important finding here: a good rating model wins plenty of races, but profit depends on testing which picks to back and at what price. Four filtered strategies made money over the same period.
Cite or share this study
Suggested citation
TrackTips (2026). Spring Carnival 2026: how a machine-learning rating model performed across 1,986 Australian races. https://tracktips.com.au/research/spring-carnival-2026
Link to this study (copy and paste into your site)
Journalists and researchers: email support@tracktips.com.au for the underlying aggregate figures or a comment.
See every runner rated, every race day
TrackTips rates every runner in every Australian race, pins our top 3 above each race, and lets you backtest strategies on real results.
Try every track free for 7 days
Card required for the trial; cancel any time before day 8 and you won’t be charged. 18+.