What has shipped, what is being worked on, and why. Balance changes and pool changes get
written down here rather than landing silently — if the odds of a perfect season move, you should
be able to find out when and by how much.
Shipped
Three data defects fixed, and the odds re-measured
Forty-three players could not be drafted at all. Their names arrived with HTML-escaped apostrophes — N'Golo Kanté among them — so tapping their card silently did nothing, and the fame floors that should have protected their ratings never matched their names either. Fixed at the source: every card in the pool can now actually be picked.
Twelve more famous names — Modrić, Ibrahimović, Čech and Kovačić among them — were spelled with accents in the fame table and without them in the club feed, so their rating floors reached their national squads and missed every club card they had. Modrić's Real Madrid 2017 card was rated 70; with fame finally arriving it reads 88. The matching now derives these folds from the data itself, and refuses the ones that would be wrong — Gérson of Brazil 1970 does not lend his reputation to the Gerson playing for Marseille in 2022.
Goalkeeper ratings on club cards described nothing. The source data has no saves or goals conceded in it, so a club keeper's defence number was mostly a constant — the match engine could prefer a reserve to Keylor Navas. Club keepers are now rated from their own overall, exactly the way World Cup keepers always were, and the defensive calibration was re-fitted to the corrected data.
Two of the draw-protection rules turned out to be wired to a stale reference and had swapped fates: the star floor never ran, and the keeper-and-defender quota ran on every late spin whether you needed one or not. Both now do their jobs. The practical difference is that a draft can no longer hand you four spins in a row with nothing worth taking.
All of this makes squads genuinely stronger, so the odds of a perfect season moved, and we re-measured them at thirty thousand drafts per pool as the method requires: 1 in 3,639 careless and 1 in 120 careful in European league mode; 1 in 1,746 and 1 in 47 in English top flight. Easier than the pair published on 9 August — because a dozen of the most-drafted cards in the game just received the ratings they were always supposed to have.
The tier ladder is re-based to the new season distribution for the same reason, so CHAMPIONS keeps meaning the top few percent of seasons instead of quietly drifting easier. One ladder for both club modes, English landing a point or two higher on it — published, as ever, rather than tuned away.
Shipped
English top flight is live, and the club pool goes back to 2015
The third competition is in the game. English top flight mode draws only from English clubs — 35 of them, 220 squads, 2015 to 2025 — with its own leaderboard alongside European league and World Cup.
Getting there meant going back for three seasons that existed and had never been fetched. Every club pool now starts at 2014/15 instead of 2017/18, so European league mode grew too: 165 clubs and 1,074 squads, up from 149 and 780. That is where Leicester's title season comes from, and Hull, Middlesbrough, QPR and Sunderland with it.
One consequence worth stating plainly: ratings are percentiles across every club season in the game, so widening the pool re-ranked all of it — a player you remember as a B+ may now read B, with career arcs re-derived to match and not a single number hand-adjusted.
It stops there, and not for want of trying. Every rating in the game is derived from shot-level expected-goals data, and that data does not exist publicly for earlier seasons — it was never recorded. Going back further would mean a second, incompatible rating system sitting in the same pool as this one, which is not a trade worth making.
Shipped
A club can now come round twice in a draft — just not in a row
The old rule capped a club at three appearances per draft in European league mode and one in World Cup. The cap is gone in both club modes. A club that has just been shown sits out the next three spins instead, with no ceiling on the total.
This is a real balance change to European league mode, not just a setting for the new one, so it is written down here rather than landing quietly. A lucky run can now give you five or six squads from the same club across eleven spins. It is less dominant than it sounds, because you can never take the same player twice — a second visit is a second choice from what is left, usually of a different season.
World Cup is untouched. Eighty-eight nations is far too thin a pool to let any of them repeat.
Shipped
The odds of a perfect season, re-measured
The pair we published in the FAQ — 1 in 67,125 for a careless draft, 1 in 155 for a careful one — did not reproduce, and we said so at the time rather than leave it standing. Here is why, and what replaces it.
The odds of an unbeaten season are the per-match win probability raised to the thirty-eighth power. That is a tail statistic: at the thousand-odd drafts the original run sampled, one exceptional draft carries the whole average and the answer swings by multiples between runs. The same pool measured at 1,000, 8,000 and 30,000 drafts gave 1 in 6,760, 1 in 17,221 and 1 in 29,917. Only the last is steady enough to print.
So, at 30,000 drafts each: 1 in 29,917 careless and 1 in 130 careful in European league mode; 1 in 9,249 and 1 in 102 in English top flight. English runs easier because the great sides are a much larger share of 35 clubs than of 165. The harness now prints a confidence interval beside every figure, so this particular mistake is harder to make twice.
In the pipeline
English top flight is coming as its own mode
A third competition alongside European league and World Cup, drawing only from English clubs. Same eleven spins, same thirty-eight games, same impossible target — but every squad in the draw is one you have watched.
It gets its own leaderboard. A score set in a pool of thirty-odd clubs is not comparable to one set across five divisions, and pretending otherwise would make both boards meaningless.
There is now a proper FAQ covering how the draft works, where the ratings come from, and what the A++ to F-- grades actually mean.
It also put numbers on something we had only ever described: how much better a careful draft does than a careless one at going unbeaten. The answer is hundreds of times over.
Those figures did not hold up, and the corrected ones are in the 9 August entry above. The short version: the estimate needed thirty times as many drafts before it settled.
