Our football picks this year stand at 29 wins from 48. That is 60%, and we publish it, and we also say plainly that it proves nothing — because at forty-eight bets the range of outcomes a coin-flipping analyst could produce comfortably includes it. Here is how to read a record, yours or ours, without fooling yourself.
How much noise there is
At even money, a bettor with no edge at all wins 50% on average, but over fifty bets the ordinary range is roughly 40% to 60%. Over two hundred it narrows to 45% to 55%. Over a thousand, to 47% to 53%. A genuine 3% edge — which would be excellent — is invisible inside the noise until several hundred bets have settled. Anything you conclude before that is a guess with a spreadsheet attached.
What to measure
- Return on stakes — profit divided by total staked, in units. £40 profit on £1,000 staked is +4%; the pounds figure on its own says nothing about the volume behind it.
- Closing line value — the average gap, in probability points, between the price you took and the closing price. It stabilises far faster than profit does.
- Calibration — the average of the probabilities you wrote down against your actual strike rate. If you said 58% on average and won 51%, the market was right and you were not.
Do not read the record by month
A month is twenty bets. Every month will look like a hot streak or a slump, and neither is information. Read the record at a hundred bets, then every hundred after that, and change nothing in between except what you had already decided to change.
What the numbers can tell you
Positive CLV and a losing hundred: keep going. Negative CLV and a winning hundred: the win is luck and the CLV is the forecast. Calibration off by more than five points in either direction: your probability estimates are systematically wrong in a way you can now correct. And if all three are flat after five hundred bets, the market knows what you know, which is the normal condition of most bettors and no disgrace — it just means the only edge left is the price, and the line shopping guide covers that.
