Over the Odds Back to the board

Our rules

Everything the board does, written down. No secret sauce, because the rules aren't the clever bit — doing them the same way every single day is.

If you want the idea behind it rather than the settings, start here instead.

What gets on the board

Twice a day we collect prices from every UK bookmaker we can reach, work out what the market as a whole thinks each outcome is worth, and list anything priced generously against that. A bet only appears if it clears all of these:

Rule	What it means
At least 6 bookmakers	If fewer than six firms have priced a market, we ignore it. Two or three bookmakers aren't a market, they're a rumour. Handicaps need eight: we found that thin handicap markets could never be checked at the off, so we were publishing bets we had no way of marking.
At least 2% better	The price has to beat what the market says by this much. Anything smaller is within the noise.
Nothing above 6.00	Long shots carry enormous margin and the market agrees with itself least on them, so the maths is least trustworthy exactly where it looks most exciting.
Nothing above 12% either	This one surprises people. If our sums say a firm is offering a fifth more than the price should be, that isn't generosity, it's a mistake at our end — usually we've compared two things that aren't quite the same bet. Sixteen bookmakers watching the same match don't all get it that wrong together. So anything that good gets binned rather than published. It's a smoke alarm, not a filter.
Starting in over 2 hours	Prices move fastest near the off. If we can't give you a fair chance of still getting it, we don't list it.
Same line only	Over 2.5 goals at one firm and over 3.5 at another are different bets. We only compare firms offering the same thing.
One row per bet	Where several firms offer the same selection, we show the best price and list the others underneath, rather than filling the page with the same bet.

What we cover

Tennis, the NBA, the NHL, the NFL, cricket, darts and snooker.

Not football. It was on the board until 26 October and came off because our own record showed no edge there: a coin flip against the closing price across 128 checked bets. The record page keeps every football selection we published and shows the numbers by competition.

Where the "market price" comes from

Every bookmaker's prices add up to more than certainty, and the extra is their cut. We strip that out of each firm's prices separately, which leaves their honest view of the chances. Then we take the middle of all those views.

Two details that matter. We leave a firm out of the calculation when we're judging that same firm, or a generous price would drag the average towards itself and hide the thing we're looking for. And we use the betting exchanges in the calculation, because their prices are sharp, but we never list an exchange as a bet — they charge commission and plenty of people don't hold an account.

How we check whether any of it works

Between about 45 and 3 minutes before each event starts, we go back and look at what the whole market has settled on. That's the fairest test available: did the price we published beat the price the market ended up at?

We take the closest check we manage. If the market had already closed, or too few firms were left pricing it, we record that the check failed and leave the bet out of the figures entirely. We never estimate a price we didn't see.

This is a better early test than profit. Profit over fifty bets is luck. Beating the market's final price, consistently, is the thing that eventually turns into profit.

How results are settled

Outcome	How we count it
Won	Profit at the price we published, on a notional £10.
Lost	Minus the £10.
Push	Stake back. It counts as neither a winner nor a loser, exactly as a bookmaker would treat it. Lines like over 2.0 goals finishing 1-1 do this.
Anything else	Left out of every figure and flagged on the record page as unrecognised, rather than quietly counted as a loss.

Results come from the same data provider as the prices, and nothing is settled by hand.

What we never do

We don't remove a bet from the record because it lost. We don't change the price we published after the event. We never delete a bad month, a bad sport or a bad week from the record. Every selection that goes on the board ends up there, win or lose, and stays there.

Stopping is not the same as deleting. We stopped publishing football in October because 235 selections of our own record showed no edge in it. All 235 are still there, because they are the reason.

If we change how any of this works — a different threshold, a new sport, a change to the sums — the date and the settings are stored against every bet, so the before and after can be compared rather than blended together.