TennisEdge's AI model finds no betting value in L. van Assche vs A. Blockx (Estoril on 2026-07-26) — the market is efficient, with A. Blockx favoured at 1.60.
ESTORIL FINAL — L. van Assche vs A. Blockx (clay) MODELS vs MARKET - Betfair ML: van Assche 2.62 (38.2%), Blockx 1.58 (63.3%). Book 2.35/1.60 is stale context; exchange is truth. - ELO: Blockx 67.4% — agrees with the exchange. - ML model: van Assche 56.1% — a LONE split against ELO + market. - Blended (ELO+ML) van Assche ~44% vs 39.4% effective implied (2.62 comm-adj 2.54) → ~4.6% raw, below the 5% bar, and it exists only by trusting the lone ML model over ELO. - ELO/ranking sanity: Blockx rank 36 with strong ELO — credible. Van Assche rank 97 with modest ELO — consistent. No divergence flag. RECENT FORM (5 most recent each) Van Assche (rank 97, age 22): Estoril run W4 — SF d. Gaston (1696) 6-4 7-5 DOM-vs-Strong; QF d. Rublev (1959, Elite) in 3; R16 d. Carreno-Busta (1837, Strong) in 3; R32 d. Ferreira Silva (1505) in 3. Prior: Umag R16 loss to Droguet. Hot week, but three of four wins were 3-setters. Blockx (rank 36, age 21): Estoril run W3 — SF d. Darderi (1937, Elite) 5-7 7-6 7-6 GRIND; QF d. Burruchaga (1813, Strong) in 3 GRIND; R16 d. Jacquet (1619) in 3 with TB. Prior: Umag R32 loss to Droguet; took Zverev to 4 TB sets at Wimbledon. Class resume (Ruud, FAA, F. Cerundolo scalps in Madrid). FATIGUE (precomputed) - Van Assche: 6 matches/14d, 4 in 7d, 18 sets, 938 court min, b2b 2, last 07-25. - Blockx: 4 matches/14d, 3 in 7d, 12 sets, 713 court min, b2b 1, last 07-25. - Court-time diff 225 min > 60 → van Assche disadvantaged. Age modulation: van Assche is 22, recovers fast → softened one tier to "slight edge Blockx". Both played semis yesterday. NARRATIVE - Let-down: van Assche won Parma Final 36d ago — no letdown window. Blockx none. - Venue momentum: both undefeated this week at Estoril (it's the final). Neutral. - Surface transition: none — both fully on clay. ALT MARKETS - Handicap: Blockx -3.5 @2.18 (45.9% implied), sim cover 52.8% → 6.9% edge, but laying -3.5 is HIGH-VARIANCE (bar ≥10%) → fails. Margin evidence also negates: Blockx's three wins here are all grinding 3-setters (two decided in TBs) — his scoreline distribution is tight, not blowout. Van Assche +3.5 @1.78: sim 47.2% vs 56.2% implied → negative. - Total 22.5: under @1.93 (51.8% implied) vs sim 54.1% → 2.3% edge, thin; qualitatively both players live in 3-setters, contradicting the under. Pass. GATES G1 edge-origin: FAIL — no numerical edge from model-market disagreement absent adjustments: blended ~44% vs 39.4% effective = ~4.6% < 5%, and it rests entirely on the lone ML split vs ELO+exchange. G2 consensus: FAIL — backing van Assche means fighting exchange + ELO on a single model signal; Blockx's live this-event evidence (Darderi 1937, Burruchaga scalps) sits against the pick, no hard independent evidence for A. G3 fatal-risk: FAIL — A-side thesis is negated by his own load (938 min / 18 sets / 6 matches in 14d, softened but real); B-side price (1.58, 64.5% effective implied) exceeds any honest Blockx probability (~58-60%). VERDICT: PASS. Models split, exchange efficient at both prices, no alt-market edge clears its bar. Match-winner forecast: Blockx (class, lighter week, ELO + market consensus).