How to use this platform

The single most useful stat in modern football. Every shot is given a probability of becoming a goal based on where and how it was taken — distance, angle, header vs foot, one-on-one vs a crowd. A tap-in from the six-yard box might be worth 0.7 xG; a hopeful 30-yard strike 0.03 xG.

Add up a team’s shots and you get how many goals they “should” have scored from the chances they created. If a team wins 1–0 but is out-chanced 0.4 to 2.1 xG, the scoreline flattered them — they rode their luck. Over many games, xG predicts future results better than goals do, because finishing is streaky but chance creation is repeatable.

The same idea applied to the pass before the shot. A through-ball that sets up a clear chance earns high xA even if the striker misses. It rewards the creator for the quality of the opportunity they made.

A picture of where every shot was taken. Bigger dots = higher xG (better chances); colour marks goals. A cluster of big dots in the box means a team is creating clean looks; lots of small dots from distance means they are settling for low-percentage shots.

A single number for how strong a team is, borrowed from chess. Win and you take points from your opponent; the bigger the upset, the more points move. The gap between two teams’ ELO translates directly into a win probability, which is how we resolve every simulated match. Roughly: 2000+ is elite, 1800 is solid, 1500 is a minnow.

Our blended strength score combining attack and defence quality into one figure, used to rank teams on the Rankings page. Higher is better.

The string of recent results (W / D / L), most recent last. A quick read on momentum, but a small sample — three good results can hide a weak underlying performance, which is where xG helps.

Rather than guess one outcome, we play the entire rest of the tournament 8,000 times. Each run completes the groups, seeds the bracket, and plays out every knockout tie using the teams’ win probabilities — with a dose of randomness, so upsets happen just like in real life. We then count how often each thing occurred. That count is the probability.

The share of those 8,000 simulations a team won the whole tournament. Title 18% means they lifted the trophy in about 1,440 of 8,000 runs. It already accounts for how hard their likely path is.

How often a team reached each round — the Round of 16, quarter-final, semi-final, and final. These only fall as you move right (you must reach the semi before the final), and the drop-off shows where a team’s run is most likely to end.

The chance of finishing top of the group versus merely qualifying. Winning the group usually means an easier knockout draw, so a high advance % paired with a low group-win % flags a team likely to take the hard road.

How a team’s title chance has moved since before kick-off. A green + means they’re over-performing the pre-tournament market; red means they’ve disappointed. It’s the story of who is rising and fading.

The race for top scorer. We take each player’s goals so far, fold in their xG (so a striker scoring from thin chances isn’t expected to keep it up — and an unlucky one is expected to bounce back), and project a final tally. “Win Boot” is how often they finished top scorer across the simulations.

What a bookmaker’s odds say the chance is. Decimal odds of 4.0 imply a 1 ÷ 4.0 = 25% chance. Comparing this to our model’s probability is the whole game.

The bookmaker’s built-in margin: add up the implied probabilities of every outcome and they total more than 100%. That extra is the house edge. “De-vigging” strips it out to get the market’s true estimate, which is what we compare against — a fairer fight.

Edge is the gap between our probability and the market’s. EV turns that into an average profit-or-loss per unit staked if the bet were repeated forever. Important: a positive edge almost always means our model is wrong, not that there’s free money — the market is very sharp.

A formula for how much to stake given your edge and the odds — it grows the bankroll fastest in theory while avoiding ruin. We show a fractional (conservative) Kelly, because full Kelly is famously wild. Shown for discipline and illustration, not as a recommendation to bet.