Whatever you think about gambling, betting odds are the most rigorously tested prediction system in football. They're produced by organisations with a direct financial incentive to be accurate, adjusted continuously by market participants, and evaluated against outcomes thousands of times a week.
As a reference point for what's likely to happen, they're hard to beat. Worth understanding what they're actually telling you.
Converting odds to probability
The basic mechanics. Decimal odds of 2.50 correspond to an implied probability of 1 divided by 2.50, or 40%. Fractional odds of 3/1 correspond to 1 divided by 4, or 25%.
Do this for all three outcomes of a match and you'll find the probabilities add up to more than 100% — typically 104 to 108% depending on the market and the bookmaker.
That excess is the margin, sometimes called the overround or the vig. It's how the operator makes money, and it means the implied probabilities are systematically inflated relative to the bookmaker's genuine estimate.
To get a cleaner probability estimate, divide each implied probability by the total. If the three outcomes imply 45%, 30% and 30%, summing to 105%, the adjusted estimates are roughly 43%, 28.5% and 28.5%.
Why they're accurate
Two mechanisms.
Bookmakers employ modelling teams with access to extensive data, and their prices start from a model estimate. That estimate is generally at least as good as anything published publicly.
Then the market adjusts. Money coming in on one side moves the price, incorporating information the bookmaker didn't have — a team news leak, a professional bettor's model, local knowledge. Prices at kick-off are meaningfully more accurate than opening prices for this reason.
The result is an estimate that aggregates a lot of independent information. That's a genuinely powerful process and it's the same reason prediction markets work reasonably well in other domains.
The known biases
They're not perfect and the imperfections are reasonably well documented.
Favourite-longshot bias. Long odds tend to be worse value than short odds — outcomes priced at 20/1 win less often than 1 in 21 times. This has been observed across many sports and markets.
Popular team bias. Heavily supported clubs attract sentimental money, which can push their prices shorter than the true probability. Prices on smaller clubs are often more accurate for this reason.
Overreaction to recent results. Markets can move too far on a run of form, for the same regression reasons that affect everyone else.
These biases are small. They're not large enough to make betting profitable after margin for most people, which is worth stating plainly.
Using them without betting
The genuinely useful application, and the one I'd recommend, is as a sanity check on your own reasoning.
If you've convinced yourself that a team is certain to win and the market has them at 55%, that gap is information. Either you know something the market doesn't — possible but unlikely — or you've overweighted something.
It's a good discipline for calibration. Anyone who writes predictions should periodically compare their confidence to market prices, because it reveals systematic overconfidence very quickly.
The market is also useful for understanding what a result means. A team beating opponents priced at 3.00 has done something more impressive than beating opponents at 1.30, and the odds give you a rough quality-of-opposition scale that league position doesn't.
What the movement tells you
Price movement between opening and kick-off is informative in itself.
A significant move usually reflects information — team news, weather, or professional money based on a model. Following the direction of a large move is more informative than following the opening price.
Very late moves, in the last hour, are often team news related, since confirmed lineups arrive around then. A team drifting sharply an hour before kick-off has probably lost somebody important.
The obvious caveat
One further use that costs nothing and is genuinely informative: comparing prices across several operators. Where they agree closely, the market is confident and there is little disagreement about the likely outcome. Where they diverge noticeably, somebody is uncertain, and that uncertainty is itself a signal that the fixture is harder to read than it looks.
Divergence tends to be largest in matches involving teams that have recently changed manager, teams returning from a long break, and fixtures where team news is genuinely unclear. Those are precisely the matches where any prediction, including yours, should carry wider error bars. The spread of prices is a free measure of how much anybody actually knows.
Odds being a good prediction does not mean betting is a good idea. The margin means the average bettor loses money over time, by construction.
The mathematics here is not ambiguous. A market with a 5% margin returns 95 cents per dollar wagered on average, and no amount of football knowledge changes the structure of that. Beating it requires being more accurate than a market that aggregates professional modelling and informed money, which is possible for a small number of people and is not the typical experience.
There's also the well-documented harm associated with gambling, which sits uneasily with how thoroughly it's integrated into football coverage now. Using odds as a reference point costs nothing. Acting on them is a different decision entirely, and the odds themselves will tell you it's not one that pays.