Visuals

Visualizations we build from our own shot and model data. No third-party embeds, just the numbers behind the leagues we cover, drawn so you can read them at a glance.

Understat · per player Shot maps Every shot a big-5 player took this season, placed where it was struck and sized by expected goals. Where a striker lives, and how good his chances are. Open → Understat · per player Finishing & method Goals against expected goals, so you can see who is running hot and who is due. Plus how each scorer gets there: open play, set pieces, penalties. Open → Model · per player Player archetypes Every attacker plotted by how much they score against how much they create. The scorers, the creators, the dual threats, all in one map. Open → socceraction · per player Action value (VAEP) Every on-ball action valued by its effect on scoring and conceding, summed per player. Who really drives play, beyond who finishes it. Open → Understat · players + teams Set-piece targets The aerial and dead-ball threats to back, and the team attack-vs-defense mismatches to target. Set pieces are where prices lag reputation. Open → Embedding · t-SNE Style map The learned style universe: every attacker placed by their per-90 profile so similar players cluster. Search a name to find them. A visual companion to comparables. Open → Model · style similarity Player comparables Pick a player and see who plays most like them across every league. Scouting a replacement, or reading a prop by a player you know better. Open → Simulation · per match Edge engine One Dixon-Coles simulation prices the whole board for any fixture: 1X2, over/under, BTTS, team totals, Asian handicap, correct scores. Fair American odds. Open → LightGBM · in-house Our xG model Expected goals built in-house and cross-validated on 535k shots. The pitch value surface, the calibration, and how it stacks up against the vendor number. Open → our xG · fade / back Regression watch Off our own xG: who is finishing above their shots and due to cool (fade), and who is ice cold but getting chances (back). The mean-reversion angle. Open → Monte Carlo · per team Season projection The whole season simulated thousands of times into a projected final table, with title, European and relegation probabilities. The outright markets, priced by our model. Open → Bayesian · per team Team strength A Bayesian strength rating that knows how sure it is: partial-pooled attack minus defense with a confidence interval, and a Kelly stake multiplier from that certainty. Open → Elo · per team Power rankings Our own Elo from 12 seasons of results, margin-of-victory scaled. One clean strength number per team, ranked within a league and across all of them. Open → Understat · per team xG table The league table it would have been on the underlying numbers. Expected points beside real points shows who over-performed their shots and who got unlucky. Open →