NFL

Leverage Index

Every game gets a 0 to 100 score for how much the result changes each team's season. For the NFL, NBA, NHL, MLB, Premier League and Champions League, the board simulates the rest of the season 2,000 times from current standings, once with each team winning. The swing is a team's odds if it wins minus its odds if it loses, added across both teams. What counts as the prize follows each league: NFL division titles plus wild cards, NBA top-6 seeds and the play-in, the top 8 per NHL conference, the top 6 per MLB league, the Premier League title, top 4 and relegation, and the Champions League top 8 and top 24.

Playoff series use series math: a Game 7 scores near 100, a 3-0 lead scores far lower. College football and college hoops use a simpler estimate built on record, AP rank and conference.

Future games are scored on today's standings and re-scored every time results come in, so the number moves as the season does. The arrow on a game shows how far it has moved since you first opened it.

The 3 bets

Every pick on the site is made once, on our server, and everyone sees the same ones. The server re-runs the model every 15 minutes. From the morning on, each game shows the 3 bets the model is leaning toward. They lock 1 hour before that game starts. The game bet compares ESPN's matchup model (football and basketball) or our record-based rating blended with the market (hockey, baseball) to the DraftKings price. Soccer uses a goals model from recent scoring to price the result, total goals and both teams to score. No preseason games, no underdog moneyline unless we have it close to a coin flip, and every game bet needs at least a 40% chance.

Our chance on every play is a blend: 60% our model, 40% the market's price with the vig removed. When a model disagrees with the book by a mile, the model is usually the one that's wrong, so big gaps get pulled back toward the market. That keeps the edges honest.

After a pick locks, the server keeps checking injuries and probable pitchers until the game starts. If the player is ruled out, the pick is swapped for the next best play and the swap is shown with its time. If it's too late to swap, the pick voids, like at a sportsbook. During the game each pick shows whether it's on track, and after the final it's graded as a hit, miss, push or void. That's the official record, the same for everyone.

Pick of the Day and parlays

The Pick of the Day is the single biggest edge of the day: overs and yes-plays only, priced -200 to +150, at least a 55% chance, no long shots. Brown's Boomer is the reach (plus-money legs), Brown's Underwear is the safe one (60%+ legs). Both use one leg per game. They lean all day and lock 1 hour before the day's first game on weekdays, or at 7:15 am ET on weekends. Each league page also has its own Pick of the Day and a 3-leg parlay (55%+ legs, 3+ games on the slate), locked 1 hour before that league's first game. Game bets only make these when we're very confident: a winner needs a 70%+ chance at -200 or better, a spread or total needs 62%+.

NHL shots: for every game on the slate, the board pulls each team's box scores from this season only (up to the last six games) straight from the NHL's official stats feed. Players only show up for the team on their current NHL roster. If someone is traded mid-season, his games with his old team come with him, marked with that team's code and counted at half weight, and they fade out as he plays more for his new team. Early in the year, when a player has just a game or two, his projection leans toward the book's number (or a typical skater at his position when there's no line) until his own numbers build up. Overtime games are scaled back to 60 minutes, recent games count more than older ones, and the projection is nudged by how many shots the opponent allows. That gives each player a chance to clear 2+, 3+ and 4+ shots. When DraftKings has posted shot lines, the edge is our chance minus the book's. Before that, each play shows the fair price to beat. Plays that would cost -300 or worse are left off. Blocked shots run through the same engine with a toggle on the board: blocks per game this season, adjusted for how much the opponent shoots, with plays from 1+ to 3+ (4+ only for heavy blockers).

Player props for every other league use the same method. For the NFL and college football (passing, rushing and receiving yards, receptions, completions, passing TDs, anytime TD), the NBA and college hoops (points, rebounds, assists, threes, PRA), MLB (pitcher strikeouts, hits, home runs) and soccer (shots, shots on target, anytime goal), the board pulls each regular's game-by-game log for this season only. It weights recent games heavier, adjusts for tonight's game environment from the spread and total, and turns that into a chance of going over or under the DraftKings line. When a line is posted, small samples lean toward the average the book's price implies, since the market already knows the player. Noisy stats lean on it much more: a hitter needs about a month of games before his own hits numbers outweigh the market's. When ESPN shows the line but not the price, it uses an estimated -120, which is close to typical DraftKings prop pricing, and marks the pick "Est." Players are tied to the team they last played for, and games with a former team count half. In the NBA and college hoops, when a regular is listed out or doubtful, his usual points, rebounds, assists and threes get redistributed to his healthy teammates in proportion to their share, counting only games they played together. Those players show a green arrow with who's out.

Season-leader props project from per-game production, the team's implied scoring and the likely game script. Prop prices aren't in the feed, so each pick tells you the line or price to take it at. Players listed out or doubtful are skipped. Best Bets are the biggest edges on the slate, Strong or Solid only.