Two ways, and both run off the same data. Do the work yourself in Raw data and Streaks, filtering by hit rate or per-90 rate until you find your own legs. Or take the shortcut: open Predictions, read the numbers behind each call, and add our top picks straight to your bet builder.

• If you want to dig. Raw data is a filterable list of every player in the fixture, sortable by hit rate or per-90 rate, so you can see the top 10 for fouls committed or shots on target and choose from them. Streaks shows who is on a run, and H2H shows who hit what in the last two meetings.
• If you want the pick. Predictions gives you the model's call on every market, with the data that produced it, and Build your bet flags our golden picks, value picks and plays where we see an edge on the bookmaker.

Either way, judge a leg on the per-90 rate rather than the name. A striker on 0.6 goals per 90 is a very different bet from one on 0.2 in the same fixture.

• Use the season rate, not the last three games. Thin samples lie.
• Check the opponent. Tackles land against sides that run at you, not sides that keep the ball.
• Check the referee for cards and fouls. Some give nearly twice as many as others.
• Check the minutes. A starter is a different bet from a substitute.

Compare the price to the chance, not to how the bet feels. A bookmaker's odds carry their margin, so the implied chance always adds up to more than 100% across a market. We price each leg from the data, strip the margin out, and show our number beside the best available price. When ours is higher, there is value. When it is not, we say so.

• A bet builder multiplies the margin by every leg you add. Four legs at a 5% margin is closer to 20% against you, which is why fewer, better legs usually beat more of them.
• Legs in the same match are correlated. Backing a team to win and their striker to score is not two independent bets, and our combined price accounts for that where a naive multiplication would not.
• Value is not the same as likely. A leg can be good value at 40% and still lose more often than it wins.

The per-90 rate for the thing you are actually betting on, and how often the player has cleared that line. A striker with eight goals looks good until you see it took 2,400 minutes. We hold both numbers for every player at 150 clubs, weighted for league strength so a Championship season is not read as a Premier League one, with thin minutes flagged so a hot three-game run does not pass itself off as form.

• Goals and assists
• Shots and shots on target
• Tackles and fouls
• Yellow and red cards
• Minutes, starts and fitness flags

Ten competitions. Seven carry full squad and player data. The other three carry fixtures, odds and predictions. Tables, squads and player numbers refresh daily through the season.

• Premier League
• La Liga
• Serie A
• Bundesliga
• Ligue 1
• Championship
• MLS
• Champions League
• Europa League
• Carabao Cup

Match markets and player props, both. Every market shows the best available price next to what our model makes it, so the gap is visible rather than implied. Build a slip leg by leg and the combined price accounts for the legs being correlated.

• Match result, both teams to score, over and under goals
• Corners and cards, team and match totals
• Anytime goalscorer, shots, shots on target
• Tackles, fouls and player to be booked

Six tabs per fixture. Build your bet is the slip builder and the community. Predictions is the model's call on every market. Streaks, H2H and raw data are the evidence behind it. The stat pack is a snapshot you can send to whoever you are betting with.

• Build your bet. Build a slip leg by leg and see our golden picks, value picks and plays, with the gap between our number and the bookmaker's shown on each. Push a slip to the community, upvote and comment on other people's, and climb the tipster leaderboard. Posting and voting need an account; reading does not.
• Predictions. What the model makes every market, from the data rather than a view.
• Streaks. Live runs for every player in the game. Who is on thirty straight matches with 1+ tackle, and whether they did it against this opponent last time. Filter by competition, player or team.
• H2H. Recent meetings between the two clubs: average goals, scorelines, both teams to score rate, and which players hit which props in the last two.
• Raw data. A filterable list of every player in the fixture. Sort by hit rate or per-90 rate to pull, say, the ten most-fouling players in the game, then take one into your builder.
• Stat pack. A shareable snapshot: the top five in each category with their last-five hit rates.

We take them off the board for you. Backing a player who starts on the bench is the most expensive mistake in a bet builder and the easiest to make, because most stats pages never update. Until team sheets are published we show every fit player, since leaving someone out on a guess is worse than including them. The moment the confirmed XI lands, every board filters itself to the players actually starting.

• Injured and suspended players come off every prediction, cup ties included.
• Boards carry a "starting XI only" marker once a team sheet is confirmed.
• Players with too few minutes for their rate to mean anything are flagged, not hidden.

Yes, and it is the newest addition. The NFL hub covers all 32 teams with game pages and per-player numbers, for building NFL parlays and player props the same way you would a football bet builder. Football is still where most of the depth sits.

Check for yourself. Every published pick is graded after full time on the recent predictions page, losses as well as wins. No model is right every time, and a tool that only shows you its winners is not one to trust.