TennisEdge's AI model finds no betting value in T. Torres vs N. Basilashvili (Estoril on 2026-07-20) — the market is efficient, with N. Basilashvili favoured at 1.25.
═══ T. Torres vs N. Basilashvili — Estoril (clay), 1/16-finals ═══
MODELS vs MARKET
Market: A 4.00 (23.8%) | B 1.25 (80.0%) [Betfair ML: 4.10 / 1.29]
ELO: A 38.6% → disagrees with market (+14.7 pts on A)
ML: A 23.2% → agrees with market
Blend: A 38.6% (ELO-driven)
RECENT FORM (last 5 shown; all fetched matches used)
Torres (A) — all ITF Futures level:
07-04 M25 Figueira SF L vs P. Schoen (1596)
07-03 M25 Figueira QF W vs Ostapenkov (1493) DOM vs Weak
07-02 M25 Figueira 1/8 W vs Gonzalez F. (1343) GRIND vs Weak (2 TBs)
07-01 M25 Figueira 1/16 W vs L. Liu (1500) GRIND vs Avg
06-27 M25 Porto SF L vs Overbeck (1600)
→ 90d sample: zero Elite/Strong wins; only Strong faced (Faria 1737) was a loss.
Basilashvili (B) — ATP level:
07-16 Bastad 1/8 L vs Tirante (1809, Strong) in 3
07-14 Bastad 1/16 W vs Choinski (1791) GRIND vs Strong (TB)
07-07 Braunschweig L vs Gaubas (1635) in 3
06-30 Milan L vs Brancaccio (1524) (ret-shaped score)
06-22 Wimbledon QF L vs Ymer (1610)
→ Erratic 2-5 stretch, but every match vs Avg/Strong ATP opposition.
FATIGUE
A: 0 matches / 0 sets 14d — well rested (16 days off; mild rust note)
B: 3 matches / 9 sets / 473 court-min 14d, last match 4d ago
→ Diff >60min: B heavier, but 4 days rest largely mitigates.
NARRATIVE
Let-down: none (B's Halle Final 36d ago — outside window)
Venue: no Estoril history for either → none
TITLE: B — Halle 36d ago (grass); Rome + Madrid finals won in May
VERDICT: PASS
The 14.7-pt ELO edge on Torres is a cross-level artifact: his ELO is built on
M15/M25 fields vs ~1500 opposition and does not transfer to an ATP main draw.
ML model (23.2%) sides with market (23.8%) — the one model tuned on features
beyond raw ELO sees no value. Qualitative read kills the edge: massive class
gap favors Basilashvili; his fatigue is real but half-mitigated by 4 days rest
and does not make a Futures player 38% to win.
Alt markets: AH A+4.5 @1.86 / B-4.5 @2.06 — no model cover anchor, and
Basilashvili's blowout-or-bust volatility cuts both ways → no independent
margin read. Total 21.5: model O 55.0% vs 51.5% implied = ~3.5% edge, below
the bar and derived from the same models. No stakeable edge in any market.