How AFL odds actually work

Decimal odds are just a probability wearing a disguise. Once you can flip a price into a percentage in your head, every AFL market becomes readable — including the bit where the bookmaker takes their cut.

Reading a decimal price

Australian bookmakers use decimal odds: the number is your total return per dollar staked, stake included. $10 at $2.50 returns $25 — your $10 back plus $15 profit. Converting to an implied probability is one division:

The margin: why both sides add to more than 100%

Take a typical AFL head-to-head: Geelong $1.65, Richmond $2.25. Convert both: 60.6% + 44.4% = 105%. Real probabilities must sum to 100 — the extra 5 points is the overround (also called vig or juice), and it's the bookmaker's structural profit. Whichever team wins, the book paid out at prices slightly shorter than fair.

To see the market's true opinion you "de-vig" — rescale each side by the total: Geelong 60.6 ÷ 105 = 57.7%, Richmond 44.4 ÷ 105 = 42.3%. That 57.7% is the number worth comparing against a model, and it's the one FootyAI uses when it calculates an edge.

Why prices move

• Money. Odds are partly a market: heavy support for one side shortens its price and drifts the other, regardless of any new football information.
• Team news. Late outs, especially key position players and star mids, move markets fast — often before casual punters have seen the news.
• Weather. Rain compresses scoring and margins; totals markets react more than head-to-head, but both move.
• Model herding. Books watch each other. A price that's out of line with the pack gets bet into line quickly, which is why "consensus odds" across several bookmakers are a more stable read than any single book.

Head-to-head isn't the only market

The same odds-to-probability logic applies everywhere: the line market (a handicap that levels the game to roughly $1.90 each way), totals (over/under a points bar), and player props (disposals, goals). Each market carries its own margin — player props typically carry a much fatter one than head-to-head, which is why prop "value" needs a bigger model edge before it's real.

The takeaway

Every price is a claim about probability, and the bookmaker's claims include a built-in fee. You don't beat that fee by liking a team; you beat it (occasionally) by having a better probability than the market's — which is a modelling problem, not a loyalty one. That's the whole premise behind value betting.