Thibaut Courtois

1997 full roster Elo (−7 from national-team results)

the ± is how settled the number is, more minutes, tighter band

Tier 2, lineups + results, how rich the evidence is, not just how much

at or near his peak, no projected upside from age

Where he starts (his team’s strength, we refuse to invent what we haven’t measured) plus two fixed nudges make the quality rating above; the last line is what national-team results have added since, taking it to his full roster Elo.

starts at Belgium's strengththe fair club-over-national anchor, every player's starting point

1959

age curve at 33 (goalkeepers' curve)the position's measured peak, no discount either way

+0

goals prevented, our post-shot modelhis percentile among minutes-qualified peers, from club seasons

+44

quality rating · club, age + measured output

2004

form: national-team results, 35 appearances3165 minutes, weighted by margin, surprise and his match stats, but shared across the XI

-7

full roster Elo · with results

1997

Where the +44 skill term comes from

Not a reputation, a measurement. We score every shot in the data ourselves and rank him among minutes-qualified peers; a fixed arithmetic turns that rank into points.

measured

0.05 goals prevented per shot on target faced

his rank

83th percentile of qualified peers

scored from

the open StatsBomb archive, our post-shot (xGOT) model

the arithmetic

46 pts/σ × z(83.3th) = 46 × 0.97 = +44

A forward-only shadow signal, as of 2026-06-27. The skill term comes from our own xG models over the open archive; the rest is earned, recorded per appearance, auditable, never back-filled. It never moves a live call.

reference feea mid-tier-club, peak-age forward of average measured skill

€8m

club tier 1923the level he plays at, the strongest signal of what a club pays

×5.03

ageage 33, past the fee-peak, value falls ~18% a year

×0.55

goalkeeper fee factorstrikers set the market's top; midfield runs close; defence and goal at the record books' discounts

×0.43

individual skillmeasured 83th percentile of peers

×1.33

our estimate

≈ €12m

fee-model v3 priors (panel-anchored, 2026-06), calibrated against our reported fees. The band is roughly two-in-three likely; contract length and release clauses are data we don’t hold. Context only, it never touches a call.