Three steps, in the order a capper actually uses them
1. Build a system
A WinPicks formula is a set of weighted statistics producing a predicted margin for every game. You choose the statistics, the weights, and the window they are calculated over. The tuner will search for better settings against eight objectives if you would rather it did the grinding overnight.
Two engines ship with it: the original power-rating model that has been in the product since 1986, and a least-squares model that fits home-field advantage from the data instead of assuming a fixed number for it.
2. Test it against forty seasons
Every result is graded against real closing lines and, where we hold them, the prices that actually traded — not a flat −110 assumption on every game. When those two figures disagree, the gap is usually the most interesting thing on the screen.
3. See which ones are the real thing
This is where WinPicks earns the money. Run a scan and you get four numbers, not one:
| Number | What it means |
|---|---|
| 9,317 | combinations the scan tried |
| 424 | the bar — how many a random sweep would throw up |
| 524 | how many came back as winners |
| 11 | the keepers — still winning on games the scan had never seen |
Those eleven are what you came for. They cleared the bar, then cleared it again on data they had never touched, and they are the ones that walk into the season with you.
4. Know if it's real in weeks, not seasons
Your record lies. The line doesn't. A season is 50–80 bets, and fifty bets cannot tell a 53% formula from a coin flip — you'd find out in March whether the thing you built in August was any good. So WinPicks grades every pick a second way: against the closing line, the market's final answer. A formula that keeps landing better numbers than the close is working — and you can see it within weeks, while there is still a season left to bet it.
How much is a better number worth? We measured it on our own graded history: one point of closing line value is worth 3.9 points of win rate in the NFL (4,384 graded bets) and 3.3 points in college basketball (65,566). Treat it like Net Units — a tuning target. Formula drifting negative against the close? Retune it now, not in March. The full closing-line data, including how fast lines settle →
What the columns mean
Every grid carries the honesty columns as well as the record. Drawdown is the worst it got along the way — a 58% record that spent the season eleven units under water is not the same bet as a calm one. Run is the longest losing streak inside it. vs rest is the same angle measured against every other game rather than against a coin, which is how you find out whether you have discovered an edge or just the league.