Bet record

Should you be betting this yet?

Not yet.

only 18 settled bets. Below about 50 the result is noise whichever way it falls.

The rule: bet when the model has beaten the market by more than luck explains — not when it claims an edge. A model's own edge estimate is a claim about itself; the only thing that can check it is a settled record. Two standard errors is the usual bar, and at these prices that means roughly 100–200 settled bets. That is not pessimism, it is what it costs to tell a real 5% edge from a coin that landed well.

You do not have to stake money to get there. Every recorded line is graded whether or not a bet was placed, so the record builds itself for free.

^* 17 rows did not come from a fitted model. Their league has too little history to fit one, so the figure is the naive midpoint — the side's own recent possession averaged against what its opponent concedes, on as few as two matches each. The midpoint pulls hard towards 50, so it understates a mismatch: a strong side away at a weak one reads lower than it should, and the edge against a bookmaker's line can come out the wrong way round entirely. Treat the Bet, Edge and Stake columns on these rows as unproven.

Adding lines

Edit odds.txt in GitHub's web editor, paste a row, commit. The hourly run imports it. Columns are date · club · club · whose-possession · line · over · under, whitespace-separated, with - for a price you do not have. Club order does not matter. Use the abbreviations below, or names without spaces.