efStats

A football match contains ~1,600 on-ball actions. Traditional metrics (goals, assists, xG) evaluate fewer than 1% of them. VAEP is the metric that values the other 99%. That's where hidden talent tends to surface. 🧵

Scout Lab — a great source for both xPass and xDribble. You can look up any other player's overperformance at the link below: 🔗 scoutlab.streamlit.app

Third: a CB whose passing skill the raw stat under-rates. Bastoni's Forward Pass% sits around the average for T5 CBs, sometimes a touch below. xPass Overperformance% says he's near the top of the position three seasons running (23/24, 24/25, 25/26). He's a wide creator, attempting chance-creating balls a typical CB wouldn't try, and completing them above the baseline for those exact conditions.

Pass completion rate doesn't care where you play, who you face, or how hard the pass was. An 80% number for a playmaker and a 90% number for a holding midfielder are not the same thing — but raw stats treat them as if they were. xPass fixes this. 🧵

@academicismo Thanks. Whatever the environment, as long as the will is there, new ideas will always emerge. Let's keep learning together.

@ef_Stats learned a lot of from you: applying Bayes here, weighting, modelling and all the little ways football can be used to teach much bigger ideas. The loss of FBref mattered in itself. But it mattered even more because it pushed people like you to keep teaching through football.

Rúben Dias similarity map using Euclidean distance Quadrants include: Compromise target = similar enough, possibly affordable Deep pockets = similar, but bring the chequebook Broken club opportunity = no need for further explanation Different profile = not really the same thing

@xenostat Highlighting is one-player-at-a-time on the site, so it gets time-consuming — faster if you check it there directly. scoutlab.streamlit.app

@ef_Stats can we see the names in 22/23 chart

Share of Team Progression vs Recoveries | T5 U-25 MF In short: the X-axis is responsibility in IP, the Y-axis responsibility in OOP. In 22/23, his final season at West Ham, Rice sat in the very top-right of this plot. And in 25/26, Elliot Anderson occupies that same spot.

@phalmer The site only lets you highlight players one at a time, so doing it for multiple players is a hassle. I decided it was only worth that effort for the latest plot.

@ef_Stats Why did you not highlight anyone else in the old chart?...

@Wanalyst007 Exactly that

@ef_Stats Elite Generalists

Just signed long-term to 2030 in January. If clubs miss out on Anderson this summer, Garner has the profile to be a realistic next option. Stylistically, Moisés Caicedo is one event-data reference point — duels, recoveries, ball-security. To climb toward that tier Garner needs more progression volume and tighter ball-security in build-up. Through the back injuries that stalled his career, one breakout season has carried him all the way to England. Wherever the ceiling is, he hasn't hit it yet. 📷: x.com/Gradient_Sport…

Comparing Garner with Anderson on Gradient grades. Anderson clearly ahead on: Passing (Anderson's progression volume is in another tier) Ball Carrying (Garner 60.5 — 84th/124 MID) Aerial Duel Garner level or clearly ahead on: Fifty Fifty Duel 87.6 (2nd) vs Anderson 72.5 Crossing 87.6 (2nd) vs Anderson 63.6 Tackling 79.1 (9th), Positioning 78.9 (3rd), Tackle Resistance 80.3 (16th) — all level or better

🏴󠁧󠁢󠁥󠁮󠁧󠁿 James Garner — Data Report Elite OOP base. Set-piece danger. Secure on the ball. And when Everton need it, he plays full-back — left or right. James Garner's 25/26 has been a real breakout. 🧵