BetterBet

Team rule lab

By · Updated · How this is measured

Loading…

Every club gets its own rules here — NFL, NBA, NRL or AFL, switch above — rather than one rule for the whole competition. For each side, every combination of 2, 3 and 4 tipping rules is tested inside that club's own games, the best is chosen on the earlier games, and it is then graded on games the search never saw.

Thousands of combinations get tried, so something will always look brilliant in hindsight. The number that matters is the holdout, and the bar it has to clear is not 50% — it is the best flat strategy over the very same games. A combination that hits 75% where "back the favourite" would have hit 76% has found nothing.

Firing this week

Loading…

The historical search

Rules are selected by search, so in-sample percentages on this page are descriptive, not forecasts. Informational only, not betting advice · 18+.

Do betting trends actually beat their own baseline?

Situational trends replayed walk-forward against the matched baseline — each team's own rate over exactly the matches the trend fired on.
CodePredictions gradedTrend hit rateMatched baselineEdge
NRL8,25155.3%57.5%-2.1
AFL18,29153.6%56.6%-2.9
NFL17,16856.5%59.6%-3

The answer measured here is no. Across all four codes, situational trends hit at or below the rate the same teams were already achieving over the same matches, so the apparent edge is the baseline, not a finding. The comparison is against each team's own rate — never 50%, never the league average — because a trend that fires mostly on good teams will look predictive against any other yardstick.

Build your own rule combination

The search above reports the best combination it found out of thousands, which means the winner is partly whichever one got lucky. A combination you name in advance is one hypothesis and one test, so the result means more. Tick rules, pick a market, and it is graded across every period at once.

No rules selected.

Conditions — these narrow which games count, without picking a side: