TennisEdge's AI model finds no betting value in M. Frech vs L. Jeanjean (Toronto on 2026-08-02) — the market is efficient, with M. Frech favoured at 1.62.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | M. Frech | L. Jeanjean |
|---|---|---|
| Elo ratings-based, surface-weighted |
66.1% | 33.9% |
| Machine learning 29-feature gradient-boosted model |
52.8% | 47.2% |
| Serve / return point-by-point simulation |
39.5% | 60.5% |
| Blend what the bot actually prices off |
66.1% | 33.9% |
| Market implied de-vigged from the opening odds |
58.7% | 41.3% |
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. Frech | hard | 1745 | 122–98 | 220 |
| L. Jeanjean | clay | 1691 | 111–70 | 181 |
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. Frech leads 2–0 over 2 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 3, 2026 | Toronto | hard | M. Frech | 2 - 0 |
| May 3, 2022 | Saint Malo | clay | M. Frech | 2 - 0 |
| W | L. Jeanjean | 2 - 0 | Aug. 3, 2026 |
| L | J. Pegula | 2 - 1 | July 30, 2026 |
| W | A. Li | 2 - 0 | July 28, 2026 |
| L | A. Kalinskaya | 0 - 2 | June 29, 2026 |
| L | N. Osaka | 2 - 0 | June 22, 2026 |
| L | M. Frech | 2 - 0 | Aug. 3, 2026 |
| W | L. Zhu | 2 - 1 | Aug. 1, 2026 |
| L | M. Sherif | 2 - 0 | July 11, 2026 |
| W | A. Tubello | 2 - 1 | July 10, 2026 |
| W | F. Jones | 1 - 2 | July 9, 2026 |
All Aug. 2, 2026 results & picks →
═══ M. Frech (A) vs L. Jeanjean (B) — WTA Toronto 1/64-finals, hard ═══
MODELS & MARKET
ELO: A 66.1% | ML: A 52.8% | Model: 66.1% (13pp model split)
Book: A 1.62 / B 2.30 | Betfair ML: A 1.58 / B 1.98
Betfair devigged: A ~55.6% / B ~44.4% (overround 1.138)
Note: Betfair 1.58 is SHORTER than book 1.62 on A — exchange erodes the
apparent book value (pricing discipline #1: soft book price the exchange
contradicts does not confirm value).
ELO vs ranking: Frech #45, Jeanjean #127 — ELO agrees with ranking
direction; credible, no farmed-rating flag. But ELO (66%) is the OUTLIER
vs ML (53%) and market (56%).
RECENT FORM
Frech (A) — match-sharp on hard (played Washington this week, no transition):
07-30 Washington 1/8 L vs Pegula (2083, elite) 3-6 6-3 6-0 — took a set,
bageled in the decider
07-28 Washington L16 W vs A. Li (1728, strong) 6-4 6-3 — DOM over Strong
06-29 Wimbledon L vs Kalinskaya (grass) 6-7 4-6
06-22 Bad Homburg L vs Osaka (grass) 4-6 1-6
06-16 Berlin L vs Lys (grass) TB 3-6
Shape: STABLE — solid hard win over Li, competitive loss to elite Pegula.
Jeanjean (B) — live dog, transitioned onto hard with one match in:
08-01 Toronto hard W vs L. Zhu (1679) 2-6 6-2 6-3 — WON a 3-setter yesterday
07-11 Contrexeville SF L vs Sherif (clay)
07-10 Contrexeville QF W vs Tubello (clay)
07-09 Contrexeville W vs Jones (clay)
07-07 Contrexeville W vs Monnot (clay)
Shape: STABLE — came off a ~3-week gap, one hard win to get sharp.
FATIGUE
Jeanjean: WON a 3-set hard match 08-01 (yesterday) → 3-set-hangover haircut
applies (-4.5pp, surface-agnostic). Age 30 (>=30) → hardened one tier.
This would push B DOWN / A UP — but it is a fatigue adjustment, not an edge.
Frech: last match 07-30 (2-set-ish load, ~163min/14d) — roughly neutral,
no extreme spot. Cumulative-minutes gap = context only.
NARRATIVE
No relevant let-down (Berlin grass Final 06-15 is >6wk stale).
No surface transition for Frech (hard->hard). Jeanjean transitioned clay->hard
but got a match in — treated in path read, not an ML edge here.
MARKET CHOICE
alt_markets: none at analysis time (handicap & totals null; only betfair_ml).
Stay ml. AH: not quoted.
GATES
G1 edge-origin: FAIL — pre-fatigue, A@Betfair 1.58 needs p>=66.5% to clear the
5% bar; even ELO's 66.1% falls short, ML (52.8%) and market (55.6%) say no.
Best-case edge ~1.6% (ELO) to negative (blend). Any bettable edge on A only
appears after adding B's fatigue haircut — G1 forbids fatigue-manufactured edge.
G2 consensus: stake-tier only (never blocks) — moot, no bet.
G3 fatal-risk: n/a (ml, no bet).
VERDICT — PASS
Two models disagree sharply and the exchange settles near the lower (ML) read;
ELO is the outlier and even it barely reaches break-even. No clean pre-fatigue
edge on either side; the only path to a bet is leaning ELO + Jeanjean's
hangover, which is precisely the manufactured-edge pattern the gates reject.
Jeanjean is a live, match-sharp dog, not a soft fade. No stake.