Record

Every leg a drop published, graded automatically once the game ended. Nothing is edited by hand, and every rate carries the sample it came from.

recommended legs, wins–losses
hit rate
recommended legs graded so far

What the backtest says

The NFL model was backtested on , games. It called the moneyline right of the time, which is roughly what picking the home favourite every week would give you, and it went against the spread over picks. That is below the 52.4% a −110 price needs to break even.

The live sample below is a handful of legs from a shadow run. It is far too small to mean anything yet. HourDrop is sold as analysis, not as a winning system, and no drop claims otherwise.

Recommended legs, by bucket

By tier

Bar is the hit rate. The hairline marks 50%.

Recommended by sport

Recommended by market

Every graded leg

All legs, both sides of every market

every leg of every market

Both sides of every market are graded, so this number sits near 50% by construction. It is a diagnostic, not a result.

By sport

By market

Every leg

    Edge is our probability minus the book’s implied probability, in percentage points. A leg is a win, a loss or a push — there is no third category, and none of these rows can be deleted after the fact.