Manager Ratings

This is an auditable Elo Impact baseline, not a validated estimate of individual coaching ability. Total impact measures observed improvement and can favour longer careers and repeated rebuilds. The per-38 view is volatile for short spells; no minimum sample is silently imposed. Match counts describe evidence, not statistical confidence.

Every club starts at 1,500 when first observed. Standard Elo: expected score = 1 / (1 + 10^((away rating − home rating − 60)/400)); K = 20; win = 1, draw = 0.5, loss = 0. No goal-margin weighting, seasonal reset or elite loss cap. European fixtures connect the leagues. Parameters have not yet been calibrated; initial league offsets are equal. Sparse histories, promoted clubs and opponents whose domestic leagues are absent can bias ratings.

Only completed regulation-time matches are used; matches ending in extra time or penalties are excluded consistently to avoid mistaking shootout totals for football scores. Future and unfinished fixtures are excluded. Missing managers do not prevent team rating updates, but their impact is not assigned to anyone. Missing scores or ambiguous participants exclude the fixture.

Spells are reconstructed from consecutive observed team-manager assignments. A gap exceeding 365 days starts a new observed segment. Spell dates are first/last observed matches, not verified appointment dates. A change during an unobserved competition can be missed. Warm-up-era spells are marked partial. Checkpoints show cumulative covered Elo change when 20/40 domestic league games are reached; partial spells have no checkpoints. Positive-spell counts require 20 covered matches and an observed start after the ranking window begins.

Current employment uses Sportsmonks active appointment records with head-coach/interim roles, supplemented by dated, sourced corrections for Brest and Lens. The remaining appointments have not been independently verified. Corrected appointment intervals also override stale fixture coach assignments; original provider data is retained in the source cache. Contract end dates are not treated as departures. Previous work outside covered competitions is excluded. Transfers, squad quality, injuries, inherited trajectory, regression to the mean and sustained elite performance are not separately controlled. No causal claims or predictive validation are made.

Main-ranking eligibility requires at least 20 completed, usable club matches in the rolling previous 12 months across covered competitions. Older covered matches still count toward long-term impact. Current top-five managers below the threshold appear separately. Managers outside the current top-five roster are included on their results; their last covered club is not a claim of present employment. Employment remains unverified unless explicitly stated. Deceased managers are excluded. This broader selection reduces employed-manager selection bias but does not eliminate it.

Source: Sportsmonks API. Transfermarkt is available as a backup but has not been used to amend these records. This is an independent Elo implementation, not ClubElo ratings.

Background reading: Eighty Five Points on Elo impact and the Manager Elo Database. These informed the project idea; this model uses its own implementation and Sportsmonks match data, with different scoring and eligibility rules.