Which traits & leagues predict Bundesliga success?

After we shortlist traits that travel (consistency check → drop near-duplicates), Phase 2 tests what lines up with first-season Bundesliga success. Then open player examples with before/after radar charts.

How we define Bundesliga success (Phase 2)

Our primary success marker is minutes played in the player’s first Bundesliga season (Year 1 in Bundesliga minutes). We treat more Year 1 in Bundesliga minutes as playing-time success — the club gave the arriving player a real role. Trait tests, league comparisons, and regression change in first-season Bundesliga minutes all use this outcome unless noted otherwise.

We also show Year 1 in Bundesliga minutes percentile (rank in cohort) and shortlisted-trait score (how strong the first-season profile looks on traits that travel). Transfermarkt market-value change would be the top outcome in a pro model, but we do not scrape Transfermarkt (ToS); minutes are our open-data proxy.

Method

Primary success marker: Year 1 in Bundesliga minutes — minutes in the player’s first Bundesliga season
Secondary outcomes: First-season minutes percentile · shortlisted-trait score
Trait test: Spearman ρ — prior-season trait percentile vs each outcome
League test: mean first-season Bundesliga outcomes by prior league (Football Reference, N=329)
League × trait: within each trait, which prior league’s profile best lines up with first-season Bundesliga minutes?
Regression (done · exploratory): OLS on first-season Bundesliga minutes — league indicators vs Serie A + prior minutes + traits (Big Five subset N=117)

Phase 2 includes exploratory OLS (not a validated forecast).

Overall prior leagues (Football Reference)

Descriptive means by prior league (n≥8). Success tests use final shortlists (Football Reference N=329 · 32 traits).

Traits most indicative of Bundesliga success (overall)

Football Reference (N=329)

Higher prior percentile on defensive penalty-area work, passes into the final third, shot blocks, and aerials ↔ more first-season Bundesliga minutes.

StatsBomb (N=96)

Directional only — many priors are World Cup/Euro minutes, not domestic league seasons.

Which league is most predictive per trait?

For each shortlisted trait, best prior league by Spearman ρ (prior percentile → first-season Bundesliga minutes). Split at p < 0.05.

Stronger signal (p < 0.05)

Best-league cells that clear the usual significance bar — still associative, small within-league n.

Category	Trait	Best prior league	ρ	p	n

Noisier / weaker (p ≥ 0.05)

Same ranking method, but the “best” league is still a weak or underpowered cell — do not over-read these.

Category	Trait	Best prior league	ρ	p	n

Reading ρ: Each row is the relatively best prior league for that trait, not a forecast. Top table = clearer associations (p<0.05). Bottom = highest ρ among noisy cells — “least weak,” not strong evidence.

Regression — exploratory OLS (Big Five subset)

Descriptive means aren’t enough. OLS estimates how much each Big Five league adds on first-season Bundesliga minutes vs Serie A, then asks whether prior trait percentiles explain the gap.

League effects (M1 · vs Serie A)

League	Change in first-season minutes	p

Prior traits (M3 · controlling for league)

Trait pct	Change in first-season minutes	p

Stat dictionary (glossary)

Definitions for every shortlisted trait in the player explorer, plus outcome terms used on this page.

Outcomes & methods

How to read percentiles

Prior and first Bundesliga season cells are percentiles 0–100 vs Bundesliga position peers in the cohort. 50 = average, 90 ≈ top decile. Consistency r (on explorer metrics) is prior→first-season rate correlation from Step 1 — separate from success-indicator ρ.