TennisEdge's AI model finds no betting value in D. Merida Aguilar vs L. Draxl (Montreal on 2026-08-03) — the market is efficient, with D. Merida Aguilar favoured at 1.41.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | D. Merida Aguilar | L. Draxl |
|---|---|---|
| Elo ratings-based, surface-weighted |
58.1% | 41.9% |
| Machine learning 29-feature gradient-boosted model |
62.8% | 37.2% |
| Serve / return point-by-point simulation |
60.3% | 39.7% |
| Blend what the bot actually prices off |
62.7% | 37.3% |
| Market implied de-vigged from the opening odds |
67.8% | 32.2% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| D. Merida Aguilar | clay | 1776 | 166–102 | 268 |
| L. Draxl | hard | 1669 | 166–82 | 248 |
Our own ratings, computed from match results since 2018. Players with fewer than 5 rated matches are omitted rather than shown at a provisional starting value. Last updated July 21, 2026.
| L | K. Jacquet | 1 - 2 | July 21, 2026 |
| W | D. Dzumhur | 2 - 1 | July 18, 2026 |
| W | R. A. Burruchaga | 0 - 2 | July 17, 2026 |
| W | T. Droguet | 2 - 0 | July 16, 2026 |
| W | T. M. Etcheverry | 0 - 2 | July 15, 2026 |
| L | T. Samuel | 0 - 2 | July 26, 2026 |
| W | S. Mochizuki | 0 - 2 | July 25, 2026 |
| W | C. Chidekh | 2 - 0 | July 24, 2026 |
| W | A. Holmgren | 1 - 2 | July 23, 2026 |
| W | M. Hurrion | 0 - 2 | July 21, 2026 |
All Aug. 3, 2026 results & picks →
═══ D. Merida Aguilar (A) vs L. Draxl (B) — Montreal R1 (hard) ═══ MODELS vs MARKET ELO A 58.1% | ML A 62.8% | blend A 62.7% Pinnacle no-vig A 55.2% (ml 1.758 / 2.17) | brief book 1.41 / 2.97 (stale — 1.41 implies 71%, contradicts sharp Pinnacle 55%) value_pick B/+5.1% is computed off the stale 2.97 quote only. RECENT FORM (last 5) A (Merida, 21, #86): CLAY specialist. WON Umag ATP250 title 07-18 (bt Dzumhur 1723, Etcheverry 1801, Burruchaga 1811, Droguet 1809 — genuine tour-level scalps). Lost Estoril R1 07-21 to Jacquet 1594. Also Rome clay title 05-05. Last match 07-21 (12d ago). B (Draxl, 24, #132): HARD player. Reached Winnipeg CHALLENGER final 07-26, lost 4-6 2-6 to Samuel 1851. Best hard win = Chidekh 1677; also lost to Rice 1324 at M25 Laval 07-10. Level = Challenger/ITF. Last match 07-26 (7d rest). SHAPE / PATH A: SHARPENING on clay (dominant title run over 1800-ELO field) but TRANSITION clay->hard for Montreal + 12d off = rust/level-depression risk. Higher ceiling of the two. B: match-sharp on hard (5 Winnipeg matches), but ceiling is Challenger level; lost the final in straights. Cross-current: surface transition hurts A, but A's proven ceiling (1800-ELO wins) sits well above B's (1677). Pinnacle has already discounted A from the naive 1.41 down to 55% — the surface adjustment is largely priced. FATIGUE (Step 2) A: 2 matches/14d, 311min, 12d rest — fresh. B: 5/14d, back_to_back 3, 445min, but 7d rest — recovered. No extreme spot; roughly equal. Not an edge either way. NARRATIVE A: TITLE Umag 15d ago (clay); TRANSITION clay->hard = directional risk on A, not support. No let-down flag (title 15d, not 0-4d). B: no relevant signal. MARKET CHOICE alt_markets: none at analysis time ML A at Pinnacle 1.758: +5.8% blend edge but only +1.2% on ELO; no exchange price, so book-priced 8% bar applies -> FAIL. Qualitative surface-transition read cuts AGAINST backing A anyway. ML B at book 2.97: +3.6% edge, but Pinnacle prices B at 2.17 (-8.8% edge) — the 2.97 is a stale/soft quote the sharp reference contradicts; per pricing discipline that KILLS the bet, it does not confirm it. Below 8% book-priced bar regardless. AH: not quoted GATES G1 edge-origin: FAIL — no side clears its bar at a credible executable price (A 5.8% < 8% book bar and surface-discounted; B negative at sharp Pinnacle, positive only off a stale 2.97 quote). G2 consensus: PASS (stake-tier only) — moot given G1. G3 fatal-risk: n/a — ML candidate, not a margin bet. VERDICT: PASS. Models (58-63% A) and sharp Pinnacle (55% A) broadly agree A is a moderate favourite; the real surface-transition concern is already priced by Pinnacle's discount off the naive 1.41. No exchange market exists, and neither side offers a clean numeric edge at a price a subscriber can actually take. No bet.