How it works

How to read a bet

Puka Nacua · Over 95.5 rec yds · bet if line ≤ 98

1. Over 95.5 rec yds is the bet and the sportsbook line we saw.
2. Bet if line ≤ 98 is the worst number still worth taking. Open your sportsbook and find his line. If it has moved past that number, skip it.
3. Our chance always sits beside the sportsbook’s own chance with its margin taken out, so you can see the gap for yourself.
4. No edge on a game means our number and the sportsbook’s agree, so there is nothing to gain.

Each game page shows a best bet, a safest bet and a call on the game, then the top 5 player bets with the reasons.

Where the numbers come from

A simple, open formula, not a black box. For every player it does five things:

1. Starts with his recent games (up to 6), counting most the ones where the same key teammates played as this week. If he has played fewer than 3 games this season, it fills in with last season’s games at half weight.
2. Adjusts for the defense. If this week’s opponent gives up more than average to his position, his number goes up; a tough defense pushes it down.
3. Adjusts for game flow. The spread and total tell us whether his team should score more or less than usual. Favorites tend to run more late; underdogs tend to throw more.
4. Adjusts for playing time. If his snaps went up last game, his number goes up a little; if they dropped, it goes down.
5. Adjusts for wind. In an outdoor game with wind above 12 mph, passing and receiving numbers come down and rushing goes up a little, by the amounts past windy games showed.

All the adjustments together can move his number by at most 10% up or down. The bigger the gap between that number and the line, the stronger the pick. The top 5 per game are the biggest gaps.

Safer alt lines. Instead of the main line (say, 72.5 yards), sportsbooks let you pick a lower bar like “50+ yards” that pays less. We look for a bar each player clears about 70 to 78% of the time; bars safer than that are priced too steeply. Small parlays combine the 2 or 3 safest in a game.

Spread and total points. We predict each team’s score from how many points it scores and allows and how efficient it is play by play, plus a small home-field edge. We only lean when we disagree with the sportsbook by 3 or more points, and a spread lean is shown only once the season’s graded record holds above 52.4% after 100 games; total leans are not shown (they have tested below break-even).

Where the public is betting. Sportsbooks report what share of bets and what share of money is on each side. When most bets are on one side but most money is on the other, bigger bettors may disagree with the public. It’s a clue, not a system.

Injuries. Anyone listed Out is left off every bet, and his teammates’ numbers are adjusted. Injury reports and game-day inactive lists are checked before each kickoff window. A pick on a player who is then ruled out is marked Void: it stays visible and doesn’t count as a win or a loss.

Weather. Every game shows the forecast for kickoff. Wind is built into the numbers. Rain and cold are shown but not used, because we couldn’t measure a reliable effect.

Lines and prices. When a pick shows a sportsbook name, the line and price are real and were saved at that moment. “Estimated line” means no sportsbook line was available yet.

What it doesn’t know: news after the last refresh, coaching changes, man or zone coverage, and your sportsbook’s exact line. Always check those before betting.

How it has done

Every past game was re-run using only information available before kickoff, then checked against what happened. The live record is on the record page.

Earlier tests, against easier lines

Player picks: in 2025 the top 5 per game hit 55.8% against 52.1% for picking blindly, graded against each player’s recent average. Real sportsbook lines are tougher.

Spread and total calls: in 2025 they hit 48.2% and 53.1% against the real lines. You need 52.4% to make money, so treat these as tiebreakers, not bets.

Safer alt lines hit 73.5% of the time in 2025, 2-leg parlays 53.7%, and 3-leg parlays 38.9%. Nothing is guaranteed: every extra leg makes a parlay much less likely to hit.

How much should one bet be?

A “unit” is your standard bet. Keeping it a small, fixed slice of what you’ve set aside means a bad week can’t wipe you out. Nothing you type here leaves your phone.

Where the data comes from

nflverse play-by-play, weekly player Stats, snap counts and official injury reports; ESPN for late injury news; The Odds API for sportsbook lines; the National Weather Service and Open-Meteo for forecasts. The site rebuilds itself several times a day during the season.

More detailed research tables are on the older tables page.