Seven from nine. Season up to 71.6%.
Results
The high-confidence calls mostly held up — Brisbane Lions, Sydney, GWS, Geelong and Melbourne all won as expected, several by big margins. Geelong put 122 on Collingwood at the MCG, GWS beat Essendon by 14, Melbourne handled West Coast by 32.
Two misses. The Fremantle vs Hawthorn one was always a 50/50 — Hawthorn were rated 56% but Fremantle won at Optus by 15. Fine, that’s what low confidence looks like.
The painful one was Port Adelaide. Rated 78% at home against Western Bulldogs, and they lost by 2 points. That’s the third time in five rounds Port have been on the wrong side of a high-confidence call at Adelaide Oval — either as a heavily-favoured home side or as the underdog. The model keeps reading their fixtures one way and the games keep going the other.
Technical
| Match | Predicted | Prob | Result | Margin |
|---|---|---|---|---|
| Fremantle vs Hawthorn | Hawthorn | 56% | ✗ | −15 |
| Brisbane Lions vs Carlton | Brisbane Lions | 96% | ✓ | +11 |
| Port Adelaide vs Western Bulldogs | Port Adelaide | 78% | ✗ | −2 |
| North Melbourne vs Sydney | Sydney | 86% | ✓ | +8 |
| GWS vs Essendon | GWS | 96% | ✓ | +14 |
| Gold Coast vs St Kilda | Gold Coast | 64% | ✓ | +29 |
| Geelong vs Collingwood | Geelong | 83% | ✓ | +54 |
| Melbourne vs West Coast | Melbourne | 90% | ✓ | +32 |
| Richmond vs Adelaide | Adelaide | 56% | ✓ | +37 |
Season: 58/81 (72%)
| Round | Correct | Total | Accuracy | Model |
|---|---|---|---|---|
| Opening Round | 3 | 5 | 60% | Logistic |
| Round 1 | 7 | 9 | 78% | Logistic |
| Round 2 | 6 | 7 | 86% | Logistic |
| Round 3 | 4 | 7 | 57% | Logistic |
| Round 4 | 5 | 8 | 63% | Logistic |
| Round 5 | 6 | 9 | 67% | Logistic |
| Round 6 | 8 | 9 | 89% | Phase NN |
| Round 7 | 7 | 9 | 78% | Phase NN |
| Round 8 | 6 | 9 | 67% | Phase NN |
| Round 9 | 7 | 9 | 78% | Phase NN |
| Season | 58 | 81 | 72% |
Model Notes
Port Adelaide at Adelaide Oval is now a known blind spot. Round 4: Port beat Richmond as a 21% underdog there. Round 7: Port beat Geelong as the away side (also at Adelaide Oval — Geelong were rated 79%, lost by 30). Round 8: Adelaide squeaked past Port by 1. Round 9: Port lost to Western Bulldogs by 2 as 78% favourites. The model keeps reading these games confidently and getting the margin direction wrong by enough to flip the result. The features capturing form and travel are clearly not the whole story for this venue.
Brisbane Lions and GWS were both rated 96% and won — but only by 11 and 14. Two heavy favourites in tight games. Brisbane’s win over Carlton was an upset waiting to happen — Carlton are better than their record suggests and the 96% feels mispriced in hindsight. GWS over Essendon followed a similar pattern. The model is right about who wins these games but consistently overstates the margin, which becomes a problem when one of them eventually drops.
Geelong at the MCG against Collingwood (83%, +54) was the cleanest result. Big favourite, big margin, no drama. After last round’s misadventures these are the games the model is built for.
The low-confidence call at Richmond vs Adelaide went the right way. Adelaide at 56% on the road, won by 37. The probability didn’t reflect what turned out to be a one-sided game, but at this level of confidence the model isn’t claiming to know — and getting it right is a bonus rather than an expectation.
Leave a Reply