Every one of these 197 selections was recorded and alerted before its match started.
Player, line, Sportsbet price and edge — sent the morning of the game, with the price captured at
the moment the alert went out. Nothing was added, adjusted or removed afterwards.
How it works
Before each game our model estimates how many disposals every AFL player is likely to get. It prices both sides of every line Sportsbet posts — the over and the under. When our estimate says one side is worth at least 20% more than the price implies, that side becomes a selection.
Say the line is 20.5 disposals, Sportsbet pays $1.88 on one side of it, and our model puts the chance of that side landing at 70%. That is an edge of 0.70 × 1.88 − 1 = 31.6%, so the bet qualifies.
Every selection meeting the fixed live rule from Round 11 onward is included in the figures on this page — winners and losers alike. One bookmaker, one market, one threshold. Whichever side of the line the model flags is the side we publish.
Historical results, for context
Backtested, not bet. Shown separately from the live record above.
The combined row is an arithmetic total of a backtest and a live record, not a continuous live result. All four seasons finished ahead. Two important differences between the periods: the backtest takes its price 15 minutes before the first bounce while the live selections take the morning price, and the backtest model was fitted on data that included the seasons it was tested on. Both are explained in full on the methodology page.
What you should know before subscribing
Stakes are large and variable. Every figure here uses $1,000 Full Kelly — stake = $1,000 × edge ÷ (price − 1). That averages $387 a bet and has reached $754, or 75% of the bank on one bet. Most subscribers should stake a fraction of that.
Losing runs happen. Nine consecutive losers at the worst point. The deepest fall from a peak was $9,642. Roughly two rounds in five finish down.
The live sample is short. 197 bets is encouraging, not settled. We would not ask anyone to expect +15.64% every season.
2026 Rounds 1–10 are not included. The alert system began in April 2026 and our probability archive starts at Round 11. Those earlier rounds are left out entirely — not counted as wins, not counted as losses.
Our stated probabilities run high. The model consistently says the side it has picked will land more often than it does. It stays profitable because it only has to beat the price, but treat the percentages as a ranking signal rather than literal chances.
Members receive every qualifying disposal selection — player, line, side, price and edge — pushed the morning of the game. The rule does not change between subscribers.