FairValue

Checking model

Esc
  • EnvironmentOverviewWhat FairValue measures, what it found, where to go next.
  • EnvironmentTerminalExplore, compare and track players; market, league and model views.
  • EnvironmentStoriesValuations, head-to-heads and research findings, told as stories.
  • TerminalPlayer ExplorerSearch, filter and sort the player export.
  • TerminalComparison StudioTwo to four players on one scale.
  • TerminalWatchlistsNamed collections, kept in this browser.
  • TerminalMarket LandscapeHow observed value is distributed, and where the model drifts.

↑ ↓ to move · Enter to open · Esc to closeLoading player index…

CF · Centre-Forward

Santiago

Castro

Serie A · 21 · €32M observed

RW · Right Winger

Takefusa

Kubo

LaLiga · 25 · €20M observed

Head-to-Head story

Castro v Kubo

Centre-Forward in Serie A at €32M, right winger in LaLiga at €20M. The model reads them in opposite directions against the market.

Built from Published SHAP output for the served Model A (reports/explainability); the player was in its training data.Portraits: image credits

01

On one scale

Castro€32MA €39.7M
Kubo€20MA €15.6M

observed value Model A (served, in-sample) · dashed lines: €2M, €10M, €30M tier boundaries · log scale

02

The tale of the tape

  1. 21Age25
  2. 97Appearances79
  3. 6,263Minutes, two years5,236
  4. 3,443Minutes, last 365 days1,746
  5. 75%Share of games started75%
  6. 0.30Goals per 900.15
  7. 0.17Assists per 900.14
  8. 19%European club minutes13%

Club football in the two years to each valuation date. Longer bar = larger value, not better player.

03

What moves each estimate

The input that separates them most is age: ×2.07 for Castro (21.7 yrs) against ×1.40 for Kubo (25.0 yrs).

Castro

Model A starts Castro at €2.2M and moves the estimate input by input. Age raises it most (×2.07). It ends at €39.7M against an observed €32M.

After all 20 inputs Model A arrives at €39.7M (exp 17.497). The observed market value is €32M: residual −0.216, so the observed value is below the estimate (the model over-predicts).

raises the estimate lowers it Model A estimate observed value (dashed line)Log scale: equal lengths are equal multiplications. Inputs are ordered by size of effect, not in any causal sequence. Published SHAP run, served Model A; in-sample.

Kubo

Model A starts Kubo at €2.2M and moves the estimate input by input. League raises it most (×1.83); club points per game lowers it most (×0.87). It ends at €15.6M against an observed €20M.

After all 20 inputs Model A arrives at €15.6M (exp 16.562). The observed market value is €20M: residual +0.249, so the observed value is above the estimate (the model under-predicts).

raises the estimate lowers it Model A estimate observed value (dashed line)Log scale: equal lengths are equal multiplications. Inputs are ordered by size of effect, not in any causal sequence. Published SHAP run, served Model A; in-sample.

04

Against the market

Castro: observed €32M against €39.7M, residual −0.216, so the observed value is below the estimate (the model over-predicts).

Kubo: observed €20M against €15.6M, residual +0.249, so the observed value is above the estimate (the model under-predicts).

Both estimates come from the served model, which was trained on these players: in-sample figures that sit closer to the market than an honest held-out estimate would. A residual is a disagreement between a model and Transfermarkt, not a finding that either player is over- or under-priced. Full records: Castro, Kubo.