TennisEdge's AI model finds no betting value in M. Kessler vs C. Brace (Toronto on 2026-08-03) — the market is efficient, with M. Kessler favoured at 1.09.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | M. Kessler | C. Brace |
|---|---|---|
| Elo ratings-based, surface-weighted |
79.7% | 20.3% |
| Machine learning 29-feature gradient-boosted model |
77.1% | 22.9% |
| Serve / return point-by-point simulation |
42.7% | 57.3% |
| Blend what the bot actually prices off |
70.2% | 29.8% |
| Market implied de-vigged from the opening odds |
80.5% | 19.5% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| M. Kessler | hard | 1792 | 101–54 | 155 |
| C. Brace | hard | 1542 | 93–73 | 166 |
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 Aug. 3, 2026.
M. Kessler leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 3, 2026 | Toronto | hard | M. Kessler | 2 - 0 |
| W | C. Brace | 2 - 0 | Aug. 3, 2026 |
| L | R. Zarazua | 2 - 0 | July 29, 2026 |
| L | A. Sabalenka | 2 - 0 | July 1, 2026 |
| W | O. Oliynykova | 0 - 2 | June 29, 2026 |
| L | M. Keys | 0 - 2 | June 25, 2026 |
| L | M. Kessler | 2 - 0 | Aug. 3, 2026 |
| L | Z. Sonmez | 2 - 1 | July 28, 2026 |
| L | R. Zarazua | 2 - 0 | July 22, 2026 |
| L | S. Saito | 2 - 0 | July 18, 2026 |
| W | E. Sema | 1 - 2 | July 17, 2026 |
All Aug. 3, 2026 results & picks →
=== M. Kessler (A) vs C. Brace (B) — Toronto (WTA 1000), hard === MODELS (prob A / prob B) ELO 0.797 / 0.203 ML 0.771 / 0.229 SIM 0.702 / 0.298 (model_prob — lone divergent model) Market (Betfair): A 1.10, B 4.70 -> raw B ~0.213. value_pick=B edge 0.10 is SIM-only. EDGE TEST (back B @ 4.70, commission-adj eff 4.515, breakeven 0.2215): ELO -1.9% (neg) ; ML +0.7% (below 5% bar) ; SIM +7.6% (lone). Back A @ 1.10 (breakeven 0.913): ELO 0.797 / ML 0.771 both below -> A overpriced, no value on fav either. RECENT FORM (last 5) Kessler (A): 07-29 Memphis-Q LOST Zarazua(1500) 4-6 1-6; 07-01 Wimb LOST Sabalenka(1915); 06-29 Wimb WON Oliynykova(1405) 6-0 6-0 DOM; 06-25 Eastbourne LOST Keys(1837); 06-24 Eastbourne WON Kalinina(1649) 6-4 6-3. Also Strasbourg TITLE (clay) 05-17. Legit top-50; Memphis-Q loss a blip. Brace (B): 07-28 Memphis-Q LOST Sonmez(1500); 07-22 W100 Evansville LOST Zarazua; 07-18 W75 LOST Saito(1529) 6-0 6-2 blowout; 07-17 W75 WON Sema(1356); 07-15 W75 WON Mushika(1338). ITF W75/W100 level, 3 straight Ls to sub-1550. SHAPE/PATH Kessler STABLE — tour-level results, one weak-field hard loss. Brace FADING (6-0 6-2 blowout to a 1529, 3 straight Ls) and stepping UP from ITF to WTA 1000 vs top-50. No sharpening-dog signal. B ranked 184 with ITF-inflated stats. SIM CONTAMINATION: Brace serve/return inputs built vs 1200-1600 ITF fields (opp ELOs 1500,1500,1529,1356,1338,1490,1221). Vs a top-50 those stats overstate her -> SIM 0.298 inflated by weak-field averaging (opponent-unadjusted failure mode). Correcting down erases the edge; no justified upward case. FATIGUE: B court-time 235min last match vs A 64min — cumulative context only, not an extreme spot, and argues AGAINST the sim-favoured side. Age B 21 recovers fast. No 3-set-hangover haircut. Not a betting factor. NARRATIVE: no first-Final letdown in window; no surface transition (both last on hard). Nothing amplifies a B edge. GATES G1 edge-origin: FAIL — only >5% numeric edge (B +7.6%) is a single contaminated model (weak-field sim); ELO -1.9% and ML +0.7% refute it. No clean model-market gap survives. G2 consensus: lone-model (sim only); would cap 0.5-1.0u even if G1 passed. G3 fatal-risk: n/a (ML market). MARKET CHOICE alt_markets: handicap/totals null at analysis time (only betfair_ml present). AH: not quoted ml only market. No bet. VERDICT: PASS. Value on Brace @ 4.70 is a manufactured lone-model edge — sim inflated by weak-field ITF stats, while ELO, ML and the price agree she is ~20-23%, and her fading path (blowout losses, class step-up) confirms the sim is wrong, not the market. No value on the favourite either.