Half the Clubs That Fell Never Came Back. The Championship Keeps the Rest.
A name-blind algorithm watched 28 clubs across 95 regime changes and flagged 31 major falls. Only 16 ever climbed back — Norwich in 50 months, Ipswich stuck below their old level for 282 and counting.
28 clubs, one blind algorithm, no names. How many Championship falls became permanent exile — and how many got out?
The Championship has always had an identity problem. Every season it swallows relegated sides, belches out promoted ones, and grinds 20-odd clubs into a holding pattern that can last years — or forever. We put 28 clubs through the same blind change-point detector we used for the top-flight fingerprint: find the moments the mean Elo steps to a new level and holds, with no manager names fed in. Then we checked the bounce-back data: of every major fall detected, how many clubs actually climbed back?
The Championship Fingerprint
28 clubs from the Championship tier and its PL orbit — same blind algorithm, no manager names fed in. The coloured strip shows who was in the dugout; the dashed lines are regime breaks the data found on its own. Pick a club and see when the algorithm says the real shift happened.
Method. Binary-segmentation change-point detection on the mean level of each club's monthly performance-rating series (ClubElo, 2000–present). A split is kept only if both resulting regimes last ≥24 months and their mean rating differs by ≥50 points; a merge pass then removes any near-equal neighbours. The detector is blind to manager identity. Manager / event notes are public record, attached only to a genuinely-detected point within 18 months — never the basis of detection, never invented. Rating scale: ~1300–1400 = lower Championship, ~1600 = solid PL mid-table, 1700+ = top-half PL.
The algorithm's answer: 95 regime changes across 28 clubs
The detector flagged 95 regime changes across the 28 clubs. The single largest sustained jump it found was Sheffield United, +248 Elo — a mean rating that moved from 1409 to 1658 in the regime starting Mar 2019. The detector never knew a name. It just saw the level break.
The bounce-back test
Of the major falls detected (≥80 Elo points), 16 of 31 eventually recovered to within 20 points of their pre-fall level. The fastest recovery was Norwich City (50 months). The others are still waiting — some for over a decade. The bounce-back analysis below ranks every fall by recovery speed.
The Championship Exile Test
Every major Elo fall (≥80 pts) detected across 28 clubs, ranked by how quickly they recovered to within 20 pts of their pre-drop level. No manager names were used to detect the falls or set the targets — names are attached from public record only.
Method. Recovery = months from the detected fall until the club's monthly Elo series first reaches within 20 pts of its pre-fall mean. Falls threshold: ≥80 Elo pts detected by binary segmentation. Manager names are public record only — attached after detection, never used to identify the fall. Data: ClubElo, 2000–May 2026. — Fastmaster Intelligence.
The exile question
Ipswich Town is the longest exile in the dataset — 282 months and still below their pre-fall mean, the algorithm having found no sustained recovery. These are not dramatic relegations that get corrected in a season — they're structural declines the data has watched quietly compound. The Championship doesn't just borrow clubs from the Premier League. Sometimes it keeps them.
The biggest transformations
The leaderboard ranks every detected regime change by size of Elo shift — rises and declines. Every club, every era, sorted by how much the mean actually moved.
The Biggest Shifts
Every detected regime change, ranked by size. A blind algorithm found these step-changes knowing nothing about the dugout — manager names are attached afterward, only where the public record lines up within 18 months.
Source. ClubElo weekly histories, 40 clubs, 2000–2026. Binary-segmentation change-point detection on mean level (min 24mo segment, min 50pt gap). Manager annotations: public record only, within 18 months of detection. — Fastmaster Intelligence
Follow a manager across clubs
Pick any manager to see every club-stint as a normalised Elo trajectory indexed from appointment date — the cleanest test of whether a fingerprint travels from job to job.
The Career Arc
Does a manager carry the same playbook everywhere — or is their best work a one-club story? Each line shows their Elo delta from appointment at a club. Zero = how they found it. Negative net delta doesn't mean failure — a manager who inherits a club at peak form and wins trophies can still exit at a lower Elo than they arrived. Pick any manager who appears in two or more of our 48 clubs. Tenures of 5+ months shown.
Method. Elo series from ClubElo (2000–2026). Each line starts at 0 = the club's Elo on the manager's first month. X-axis = months since appointment. Line ends at departure (dot) or continues to present. Tenures under 5 months excluded. Manager list = those appearing at 2+ clubs in the 48-club dataset. — Fastmaster Intelligence
Method. Binary-segmentation change-point detection on the mean level of each club's monthly Elo series (ClubElo, 2000–present). A split is kept only if both resulting regimes last ≥24 months and their mean Elo differs by ≥50 points. Bounce-back: for each detected fall ≥80 pts, recovery = months until the monthly series first reaches within 20 pts of the pre-fall mean. Manager names are public record only — never the basis of detection. — Fastmaster Intelligence.
