The AFL tips page publishes a margin prediction on the first match of every round. It is a single number derived from the median of every AU bookmaker's spread line. It is a good tool for office tipping comps where a margin tiebreaker often decides the week.
It is also a useful object lesson in what market consensus is and is not. On average, across a full season, the median AU bookmaker margin line beats almost every tipster alive. On any individual match, it can be off by 30 points, and it will sound confident while being wrong.
Two examples. One where the prediction was nearly perfect. One where it was not.
Example one: the Friday night blowout
Collingwood vs North Melbourne, Round 4, MCG, Friday night. The AU market had Collingwood at $1.22 and North at $4.80 in the H2H. The spread across AU bookmakers sat at Collingwood -35.5 points, with a range of -33.5 to -38.5 depending on the book.
Median spread line: -35.5. Predicted margin: Collingwood by 36 points. The Krok Odds tool is derived from exactly this median calculation across every AU bookmaker quoting the line.
Actual result: Collingwood 102, North Melbourne 65. Margin: 37 points. Prediction was off by 1.
Every office tipping comp with a margin tiebreaker was decided by whoever picked Collingwood (obvious) with a margin between 35 and 38. Probably 60% of the tipsters in the country picked something in that range because they were looking at the same bookmaker spreads.
This is what it looks like when the market is right: a tight convergence of prices across every book, a clear consensus, and an actual result that lands inside the predicted range. No magic, no insight. Just every bookmaker's trading desk independently arriving at roughly the same number, and the game obliging.
Example two: the upset that was not supposed to happen
Melbourne vs St Kilda, Round 9. Melbourne coming off four straight wins, St Kilda in the middle of a rebuild. The market had Melbourne at $1.34, Saints at $3.20. Spread across AU bookmakers: Melbourne -22.5, range -20.5 to -24.5.
Median spread line: -22.5. Predicted margin: Melbourne by 23 points.
Actual result: St Kilda 94, Melbourne 77. Margin: St Kilda by 17.
The margin-game prediction was off by 40 points on the wrong side.
This is the part people do not like. The market was unified, confident, and completely wrong. St Kilda had a surge they were not priced for. Melbourne's key midfielder went down with a hamstring in the first quarter. A couple of 50/50 umpiring calls went to the Saints. All of these things happen. Most weeks the market is right. Some weeks it is not, and when it is not, it can be wrong by 40 points.
Why the margin prediction works on average but fails individually
The market consensus margin is an expected value across possible outcomes. If you simulated the Melbourne/St Kilda game 10,000 times with the same pre-game information, Melbourne wins around 75% of those simulations and the average margin across all simulations would be something close to +23 for Melbourne. That is what the market number represents.
But the average across 10,000 simulations is not the result of the one game that actually got played. The one game that got played is a single draw from a distribution that includes some Melbourne-by-50 outcomes, some Melbourne-by-5 outcomes, and some St-Kilda-by-15 outcomes. The more rare the outcome, the more the prediction looks stupid in retrospect. But the prediction was not stupid. It was correctly centred on the expected value. The result just happened to be a tail.
AFL, like most high-variance sports, produces a fat tail of surprise results. Across a full season of 207 matches, roughly 15-20% of H2H favourites lose, and the average margin error on market-consensus predictions is around 27 points. On some games it is 1 point. On some games it is 40 points. That variance is baked in. You cannot tighten it without being a better predictor than the entire AU bookmaking industry, and almost nobody is.
How to use margin predictions sensibly
A few principles.
Use them for tipping comp margin games. Over a full season, a market-consensus margin picker will beat the average tipster by a meaningful amount. You will lose the week when the tail outcome lands. You will probably win the comp's annual margin prize regardless, because the consistency over 23 rounds compounds.
Do not bet real money on the margin-game prediction as a margin market bet. The market consensus is already priced into the margin-specific betting market. There is no edge there. You would be backing the market's own prediction at the market's own price, minus the bookmaker margin. Net negative.
If a margin prediction looks surprising, either much bigger or much smaller than you would guess, trust it. The market has seen more information than your gut has. The Round 14 prediction of Brisbane by 47 in a game you thought was a 50/50 probably reflects something real about the matchup that is not reaching your general footy awareness.
Watch the spread. The AFL tips page shows the range of spread lines across AU bookmakers alongside the median prediction. When the range is tight (say, -22 to -24), the market is unified and the prediction is high-confidence. When the range is wide (say, -18 to -28), bookmakers disagree, and the margin prediction should be treated with appropriate uncertainty.
The season-level picture
The running average error on the Krok Odds AFL margin prediction across the current season is published on the tips page. Typically it settles somewhere between 24 and 30 points of average absolute error over a full home-and-away season, roughly in line with what the AFL betting market itself achieves.
That is a useful benchmark. If you can consistently beat it with your own analysis, you have a genuine edge over the AU betting market, which is one of the harder things to do in sports. If you cannot, the market consensus is probably your best available signal for margin tipping, and that is fine.
The positive EV guide walks through why market consensus is so hard to beat. The short version: every piece of information that could matter to a margin is priced in by someone, somewhere, in the AU bookmaker ecosystem, and the median absorbs most of it. The bettor's job is usually not to beat the consensus on average, but to find the specific spots where one bookmaker's price drifts away from it.
David has been running advantage betting strategies across Australian bookmakers since 2023 and contributes long-form retrospectives, case studies, and operational pieces drawn from years of running real bets in AU markets. His writing focuses on the realities of running a sustainable AU advantage operation — what works, what fails, and the operational details most blogs gloss over.