We've rebuilt the engine behind every composite score, archetype, Scout52 Score and Find Similar match in the app. It's live now, and every player you open is scored by it.
The last update sharpened the labels. This one goes further: we pulled almost 20,000 player-seasons from 20 leagues across four levels of the game and three seasons of data, tested the old model against them, and rebuilt what didn't hold up. Some of what we found was uncomfortable. Two of the six scores were measuring nearly the same thing. Around one label in five would flip if a single score moved by a tenth of a point. More than a fifth of players were "Elite", which says very little.
Here's what's changed, and what it means when you're building a shortlist.
Seven changes that matter
Every score is judged against his own level
Each composite is now a percentile against players in the same position and the same league tier. A 5.0 is average for that level; a 9.0 is top 10%. A League Two centre-back is measured against League Two-level centre-backs, not against the Premier League, so the numbers mean the same thing wherever the player plays.
Work Rate is now Duels
The old Work Rate score overlapped almost completely with Defending, so a defender was effectively scored twice for the same actions. It's replaced by Duels: how many duels a player wins and how often he wins them. Defending now stands on tackles, interceptions and blocks, minus times dribbled past. Passing is judged against what's normal for his position, and missing pass-accuracy data no longer drags a player down.
Archetypes that hold their shape
A player's archetype is now the profile where he ranks highest against his peers, not a chain of fixed cut-offs. Labels are far steadier: about 2% change on a tiny nudge in one score, down from around 20%. Elite now means exactly the top 10% for his tier and position, and a player with no clear profile is shown as Limited rather than forced into a type.
A Scout52 Score built around his role
The Score is now 30% match rating, 50% the scores that define his archetype and 20% everything else. A Ball Winner is judged mainly on winning the ball, a Poacher on putting it in the net. It tracks match rating far more closely than before, and defenders are no longer squeezed out of the top of the rankings by attackers.
Four tiers, and honest step-ups
Leagues now sit in four tiers, with a new Tier 4 for League One, League Two and their European equivalents. For a lower-tier player the app shows what his scores look like a level or more up, based on how almost 1,800 real players actually performed after moving between tiers. A score can drop when you move him up, but it never rises. Only a minority of Tier 4 "Elite" scores stay Elite at Tier 1, and that's the point: you see the step-up before you make it.
Goalkeepers get their own model
Keepers used to be run through the outfield formulas, which never made sense. They now have their own scores: Shot Stopping, Distribution and Command, with Workload, Goals Against and Match Rating shown as context because they depend heavily on the team in front of him. Keepers get their own archetypes and age bands (Under 23, 24–29, 30–32, 33+), and in Compare, keepers are measured against keepers.
Find Similar, rebuilt
"Find me players like this one" now compares a 17-stat playing fingerprint against players in the same position, with no archetype filter hiding good matches. The match percentage is honest too: strong matches sit around 45–65%, instead of almost everyone showing 97%. In testing, it found the same player's following season among its top 15 matches two and a half times as often as the old version.
The archetypes, as they stand
Seventeen outfield profiles and four goalkeeper profiles, each with an Elite tier for the top 10% at that level.
- Central Defender
- Ball-Playing Defender New
- Progressive Defender
- Wide Defender
- Ball Winner
- Deep Playmaker New
- Box-to-Box
- Advanced Playmaker New
- Progressive Carrier
- Attacking Midfielder
- Poacher
- Target Forward
- Pressing Forward
- Inside Forward
- Creative Forward New
- Explosive Winger New
- Complete Forward
- Shot Stopper New
- Ball-Playing Goalkeeper New
- Commanding Goalkeeper New
- Complete Goalkeeper New
Retired: Build-Up Defender, Anchor 6, Creative Maestro, Workhorse and Direct Attacker. Those players now land in the profile that fits them best, most often Ball-Playing Defender, Deep Playmaker, Advanced Playmaker, Box-to-Box or Explosive Winger.
Old engine vs new
Measured on the same pool of almost 20,000 player-seasons.
| What we tested | Old | New |
|---|---|---|
| Overlap between the two most similar scores | 0.93 | 0.54 |
| Labels that change on a 0.1 nudge in one score | ~20% | ~2% |
| Players labelled Elite | 22–29% | 10% |
| Defenders in the Tier 1 top 100 by Score | 6 | 34 |
| How closely the Score tracks match rating | 0.45 | 0.81 |
| Find Similar: player's next season in top 15 matches | 6% | 15% |
Overlap and match-rating figures are correlations from 0 to 1. For overlap, lower is better.
What this means for your shortlists
Fair across levels
Compare a Championship full-back and an Eredivisie full-back on the same footing, then see what either looks like a tier up before you commit time to him.
Defenders get their due
Defending and Duels no longer count the same actions twice, and the role-based Score means a top centre-back ranks like one.
Labels you can trust
Archetypes don't flip week to week on a rounding difference, and Elite is a genuine top-10% marker again.
Saved players and lists
Anything scored under the old engine is refreshed the next time you open that player's insights or refresh the list, so older snapshots update themselves as you use the app. Early in a season, scores use the current season only, so treat low-minute profiles as a first read.
As always, the numbers get you to the right shortlist faster. The watching is still the job.
See the new engine on your players
Open any player's insights, run Find Similar, or rank a league by Scout52 Score. Free to start, with Scout from £9.99 a month.
Open Scout52 → How composite scores work