Belgium vs. Senegal - Post-Event Review

2026 World Cup, Round of 32 | Wed July 1, 2026 | Lumen Field, Seattle WA (open-air)

R32 (knockout) | Predicted vs. Actual | Calibration: ~68%

TL;DR

Result: Belgium 3-2 Senegal (AET — BEL advance to R16 vs USA/BiH winner in Seattle). 90' BEL 2-2 SEN (HT 0-1; SEN Diarra 25', Sarr 51' / BEL Lukaku 86', Tielemans 89'; Tielemans 120'+5' ET winner). BEL advance bet paid — but it required extra time, SEN out-xG'd BEL nearly 2:1 (3.54 to 1.80), and the winner came from an unpredicted starter (Tielemans) in the 5th minute of stoppage. A clinical, lucky escape, not a dominant modal win.

• Winner (advance): ✅ Belgium (~61% advance consensus) — BEL advanced (AET). The advance bet literally paid, but via the 23% draw-at-90' pathway, not the 46% win-in-90' modal. Co-modal-ish escape, not a clean favorite victory.
• Scoreline: ⚠️ draw-shape right, modal wrong — modal 90' was BEL 2-1 (g4 16% / g5 14%) or BEL 1-0 (mm3 14%); actual 90' was 2-2 draw. The 23% draw band FIRED, but the predicted draw scores were 0-0/1-1 (not 2-2), and AET 3-2 was in no model's tree.
• Who Scores: ⚠️ split — Sarr #1 SEN ✅ (51'), De Bruyne #1 BEL ❌ — Sarr (~30%) delivered exactly as predicted; De Bruyne (~28%) blanked. Lukaku (#5 supersub, ~12%) scored 86' immediately off the bench. Tielemans (not in any top-5) scored the brace + ET winner — the model's blind spot.
• Finding 11 (weather): ✅ — 20°C cool, neutral. Validated.
• Finding 12 (conservation): ✅ — 0% conservation, both full-throttle (SEN led twice, BEL chased both times).
• Finding 13 (sterile-domination): ❌ direction wrong — BEL flagged "sterile risk"; reality was BEL CLINICAL (3 goals from 1.80 xG). SEN was the better chance-creator (3.54 xG). The draw band fired for the WRONG reason.
• Finding 6 (GK premium): ✅ — Diaw (untested) was NOT the hero; conceded 3 from 1.80 xG (over-performed BEL's chances). Draw mechanism was correctly NOT GK-driven.

Calibration Breakdown: Winner 25/25 · Scoreline 5/25 · Who Scores 12/20 · Findings 17/20 · Model-breaking 10/10 · deductions 0 = 69/100 → ~68% band (C) (narrative-adjusted −2% from the 70% midpoint for the clinical/lucky win + the unpredicted Tielemans brace that decided the match).

1–5. Summary

• BEL escape via extra time after SEN led twice. Diarra (25') and Sarr (51') put SEN ahead in each half; BEL equalized late through Lukaku (86', supersub) and Tielemans (89'), then Tielemans won it at 120'+5'. The 90' draw was the honest reflection of a level match.
• SEN was the better team on xG (3.54 vs 1.80). SEN created ~5 big chances and converted 2; BEL created ~2 and converted 3 — the inverse of the pre-match sterile-domination fear. BEL won through clinical finishing + Courtois, not territorial dominance.
• Tielemans was the unforecast protagonist. Captain, every-game starter, listed in profiles as a "penalty-box-arriving runner" (g5) but ranked nowhere in any model's top-5 scorers. He scored twice including the ET winner — the model's Who-Scores blind spot for a deep-arriving DM in a sterile-bypass team.
• Lukaku supersub rule fired exactly as profiled (g4/g5: "60-70' cameo, BEL all-time WC top scorer"). On 86', scored 86' — immediate impact. The #5 secondary pick was correctly anticipated even though the #1 (De Bruyne) missed.
• Sarr (#1 SEN) confirmed as SEN's genuine threat — 4 goals in 4 WC matches (3 group + 1 R32). The hottest-scorer model got SEN right; it got BEL wrong.
• No weather, GK, referee, or lineup surprise — all model-breaking inputs were correctly profiled. Diaw confirmed, Martínez 0 pens (band held), 4-2-3-1 BEL / 4-3-3 SEN as projected.
• R32-1 deviation cap (held BEL at ~61%) was VALIDATED. The match was genuinely close (90' draw, SEN xG edge). The cap prevented over-rating BEL on squad value — exactly its purpose. A naive €549M-vs-€212M read would have over-predicted BEL.

6. Match Stats (FINAL)

Possession near-even (52/48) — not BEL dominance. xG inverted 2:1 in SEN's favor — SEN was the better chance-creator. BEL fouled nearly 2× SEN (22 vs 12) — desperate defending while chasing twice. Card count not confirmed in the brief (carried forward as UNCONFIRMED — do not resolve by guessing). Penalty count = 0, consistent with Martínez's WC26 non-interventionist form.

7. Key Learnings

1. Finding 13 (sterile-domination) was directionally WRONG here, and the draw band fired for the wrong reason. BEL was flagged "sterile risk" (0-0 vs IRN, false-9 De Ketelaere). Reality: BEL scored 3 from 1.80 xG — clinical, efficient, the opposite of sterile. SEN (3.54 xG) was the better creator; the 90' draw came from SEN's counter-attacking quality + BEL's profligacy-then-clinical-surge, not from BEL failing to create. → Refinement to Finding 13's "validate the mechanism, not the outcome" rule: when a team with a sterile flag scores ≥2 from <2.0 xG, the flag misfired — reclassify as "clinical under-dog" and do NOT credit the draw band as a sterile-validation. The draw was real; the sterile reason was not. (This is the CZE-RSA lesson recurring: right scoreline, wrong mechanism = weaker signal.)

2. NEW candidate Finding (extend Finding 7 / low-event late-winner): the deep-arriving captain-DM in a sterile-bypass team is a live Who-Scores miss. Tielemans (captain, every-game starter, "penalty-box-arriving runner" per g5's own profile) scored the brace + ET winner and was in NO model's top-5. When a favorite's attack is structured so the entire creative burden sits on one #10 (De Bruyne) who gets shadowed, the secondary scorers are the box-arriving midfielders, not the wingers. → Propose: in a "single-creator" attack (Vanaken conservative, De Bruyne shadowed), elevate the box-to-box DM (Tielemans) to the top-5 at ~8-10%, above the false-9. This generalizes the GHA-PAN "low-event late-winner from non-scorer" pattern to a knockout.

3. R32-1 deviation cap VALIDATED (extend doc-5 / the squad-value-vs-reality gap). fav_share 0.63 correctly held BEL at ~61% advance, and the match was genuinely close (90' draw, SEN xG edge 2:1). The €549M vs €212M squad-value gap does NOT translate to chance-creation dominance when the underdog has an EPL-calibre attack (Sarr, Mané, Niakhaté) + an open 4-3-3 shape. → Confirmation: the deviation cap is load-bearing for R32 favorites vs AFCON/CONMEBOL underdogs with real attacks. Do not remove it for "KO pedigree."

4. Finding 12 (conservation = 0 in knockouts) VALIDATED cleanly. SEN led twice (25', 51') and kept attacking; BEL chased both times with no stale tilt. Knockout must-win = full-throttle, as profiled. No refinement needed — this is now a multi-match-validated rule (ARG-ALG, ENG-CRO, FRA-NOR, BEL-SEN).

5. Clinical-finishing / xG-overperformance is the unmodeled variable that decided this match. BEL scored 3 from 1.80 xG (xG over-performance of +1.2). The pipeline models chance creation (xG) but does NOT model conversion variance. When a team over-converts its chances (BEL) and the opponent under-converts (SEN 2 from 3.54), the xG-loser can win. → Propose: in close knockouts (fav_share 0.55-0.65), acknowledge a ~15-20% "clinical-finishing flips the xG" scenario where the xG-loser advances — this is irreducible conversion variance, not a model error, but it should be named in the §7K tree rather than hidden in the residual.

8. R32 Bracket Progression

BEL remain in Seattle (home base) for the R16. SEN eliminated — a creditable R32 exit; they out-xG'd a Tier-1 favorite and led twice.

Post-event for TASK (BEL-SEN R32). Category 0 PASSED: BEL 3-2 AET (90' 2-2) confirmed from ESPN Matchcenter (gameId 760493) + scout brief (≥2-source gate met). Sources: sources/matches/bel-vs-sen/post-event/00-post-event-brief.md (12 categories). FINAL stats only — possession 52/48, xG 1.80/3.54, 0 pens. Card count UNCONFIRMED (carried forward, not resolved by guessing). STRICT rubric applied (Jun 30 revision): scored on the MODAL pick, no "in top-3" inflation. Flags: knockout (§7K tree validated), ET required, clinical/lucky advance flagged honestly.