100 is exactly average for that season. 112 means you scored 12% more than the typical team that year. Because it is re-based every season it survives scoring inflation: a 110 in 2015 and a 110 in 2025 are the same achievement, even though the league scored 12% more points in 2025.
Not a career table, a forward-looking one. It weights recent seasons above old ones, shrinks small samples toward the league mean, and ignores win-loss record entirely in favour of scoring, because record carries luck and scoring does not. Drag the slider to change how hard the model discounts the past and watch the order move.
In plain English: λ ("lambda") is the fade rate on old seasons. Slide it low and only the last year or two really count; slide it to 1.00 and a 2015 season counts exactly as much as a 2025 one. Nothing about the data changes — only how much weight the model gives the past.
Each season a manager played gets weight λ(2025 − year), so at λ = 0.72 the 2025 season counts 1.00, 2024 counts 0.72, 2023 counts 0.52 and so on. Those weights are applied to games played, and the weighted mean of Power Index is then pulled toward 100 by N / (N + 25), where N is the weighted game count — so a manager with one loud season cannot leapfrog a decade of evidence. Projected win rate converts the score through the same Pythagorean exponent used everywhere else, against a 100-rated opponent.
The same bump chart, re-cut by manager: one line per season, brightest is most recent. Each line is labelled with its year and the team name that year. The names change, the manager does not. Weekly game logs exist for only, so earlier seasons cannot be drawn. There is no shaded playoff band here because the field size changed between these seasons; each line's dashed tail ends at where that season actually finished.
A league average hides the interesting part. These show how spread out the league was each year, whether everyone was bunched together or three teams ran away with it.
Season totals estimate luck. Weekly scores measure it. Choose a season below and every card in this section follows it. Greyed years have no game log yet. These cards always show everyone who played that season, so the manager filter at the top does not apply here.
Standing after each week, ranked by record then points for, the same tiebreak the league uses. The shaded band is that season's playoff field: .
Each week, count how many of the other nine teams you outscored. Every week, versus everyone.
Each week's score is shown against what the site projected for it. Beating that projection regularly points to sound start and sit decisions and useful waiver work. Any single week swings heavily on luck, so the per week column carries more meaning than the season total.
Luck is actual wins minus the wins your points deserved. Pythagorean expectation says a team scoring 1,900 and allowing 1,700 should win about 8 of 14. Win 11 and you banked three wins your scoring did not earn.
These measures are built from season totals and the playoff game log, so they cover all ten seasons. The Week by Week section examines the same ground in more detail, but only for the years with a game log loaded.
Win percentage tells you where a career ended up. It does not tell you how much of it was spent winning. Games clear is wins minus losses across a whole career, the plainest measure there is. Weeks above goes finer, walking each season week by week and asking whether that manager's record was above water at the time; it covers 2021–2025, the seasons with a loaded game log. Expected vs average asks a different question again: forget the schedule, how many wins did the scoring earn above a perfectly average team?
Season length has been 13, 14 and 15 games, so the headline records are per game; raw totals are kept separately and labelled as counting records, because a 15-game season will always out-total a 13-game one. Single-season records include everyone, because one enormous year is a real record no matter how briefly someone played. The career rate tables below (win %, average finish, power index, luck) exclude one-season managers, whose tiny samples otherwise own every extreme; the career counting tables (total points, playoff wins) include everyone, since volume cannot be inflated by a short career.
In plain English: the section above shows one season at a time, so a team from a year you are not looking at cannot be found with your browser's search. This list holds every name the league has ever had, so searching for one always works.
Read across: the row manager's record against the column manager. Playoffs cover all ten seasons; regular season covers only the years whose game logs are loaded. Use the manager filter at the top to cut the grid down to the people you care about.
Every trade in the five seasons whose transaction logs are loaded. A trade counts once for each side, so the two managers in a deal each get credit for it. In and out count players, not deals: a three-for-one shows up as 3 in and 1 out for the side receiving three.
Everything below is the league's own configuration, not an assumption. The one that matters most for reading any number on this site: Deadshot is full PPR, meaning a reception is worth 1.0, double Yahoo's default. That is why scores here run 120–140 rather than 90–110, and why raw points cannot be compared against another league.
Nothing here is a black box. Each row gives what the metric answers, the arithmetic behind it, and what it cannot see.