One best role on every screen (#586)

Ross: "Ok so we have a disconnect between the best roles on player comparison and roles on the hover card", and "otherwise people are going to be confused by the different scores".

Step one of the plan on #586, computed over Manchester United, 26/27 - all 30 players. The figures were read from a copy of the live database through the app's own code: the Squad depth board, Analyse squad, every player card at every position and phase, and Best roles with the squad compared eight at a time. Nothing on this page is typed in.

The save has one attribute capture and no statistics or metrics. Attribute scores are out of 20, so they come out the same whichever players they are measured against - which means the "one group of players" part of the plan cannot show up in these numbers. It will once football is captured.

What it found

  1. The two readings name different roles in 43 answers of 122. At positions with standard options, one of them wins on overall in 26 answers of 45.
  2. Judge every role he suits within a point of his best, not only the top one. Judged on the top one alone, Rashford has no specialism out of possession - half a point below the middle as a Wide Outlet Winger - though he suits Inside Outlet Winger exactly as well and is 1st of 8 at it. A point is where the app already stops calling two scores apart. It changes four answers: Rashford, Amad and Dorgu get the role they are strong at, and Heaven is named a Wide Centre-Back from the squad's middle, the way Kim is a Ball-Playing one. Šeško and Dalot keep no specialism, because the roles they are not below the middle on are 9 and 4 points short of their best.
  3. "Below the squad's middle" is where no specialism starts, and the verdicts cannot pin it any tighter. De Ligt is 1 point below the middle on Wide and Advanced Centre-Back, which suit him exactly as well as Ball-Playing, and Kim is exactly on it for Ball-Playing. So with those roles judged, every rule gets all three verdicts right at one setting only. Judging the top role alone, "below the middle" had seemed to hold over four settings, but only because the tie-break tested de Ligt on Ball-Playing, where he is 4 below. What picks it over "the bottom 40%", which lands on the same answers, is that it needs no number of its own. At "below the middle", 13 of the 60 answers at players' filed positions say no specialism.
  4. Ties need a rule. 7 "suits best" answers change when a tie goes to the role he plays better rather than to the name. De Ligt's card names Advanced Centre-Back, 71 overall, over Ball-Playing and Wide at 75 - all three suit him at 62.
  5. Every screen moves. A different role is named in 16 of 36 Squad depth lines, 43 of 122 Analyse squad answers, 32 of 60 Best roles answers, 23 of 60 card dropdowns and 16 of 60 Roles headlines, and more keep their role but gain "no specialism".
  6. Sort and file by the best overall score, not the answer's. At centre back in possession, sorting by the answer's own score drops Kim and Maguire a place each and lifts Martínez two, because the role each of them suits asks for what he is weakest at.