TennisEdge's AI model finds no betting value in G. Diallo vs K. Jacquet (Montreal on 2026-08-02) — the market is efficient, with G. Diallo favoured at 1.57.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | G. Diallo | K. Jacquet |
|---|---|---|
| Elo ratings-based, surface-weighted |
42.4% | 57.6% |
| Machine learning 29-feature gradient-boosted model |
43.4% | 56.6% |
| Serve / return point-by-point simulation |
20.8% | 79.2% |
| Blend what the bot actually prices off |
42.7% | 57.3% |
| Market implied de-vigged from the opening odds |
60.3% | 39.7% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| G. Diallo | hard | 1733 | 122–77 | 199 |
| K. Jacquet | clay | 1612 | 85–61 | 146 |
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.
G. Diallo leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 3, 2026 | Montreal | hard | G. Diallo | 2 - 1 |
| W | K. Jacquet | 2 - 1 | Aug. 3, 2026 |
| L | T. Boyer | 0 - 2 | July 28, 2026 |
| L | L. Sonego | 2 - 3 | July 2, 2026 |
| W | B. Bonzi | 2 - 2 | June 30, 2026 |
| L | J. Draper | 2 - 0 | June 25, 2026 |
| L | G. Diallo | 2 - 1 | Aug. 3, 2026 |
| W | A. Shevchenko | 1 - 2 | Aug. 1, 2026 |
| L | A. Blockx | 1 - 2 | July 22, 2026 |
| W | D. Merida Aguilar | 1 - 2 | July 21, 2026 |
| L | T. Daniel | 1 - 2 | July 19, 2026 |
All Aug. 2, 2026 results & picks →
═══ Diallo (A) vs Jacquet (B) — Montreal R1 (1/64), hard ═══
MODELS vs MARKET
ELO 0.4243 | ML 0.4338 | sim 0.4270 -> all favor B (Jacquet) ~57%
Betfair ML: A 1.86 (53.8%) / B 2.10 (47.6%); Pinnacle no-vig A 0.5309
value_pick B, raw edge 17.5%. Large model-vs-market split: models make the
rank-136 qualifier a 57% favourite over the rank-40 direct entrant; market
makes the rank-40 man the favourite. Job = decide who's right.
ELO/RANKING SANITY (decisive)
B ranked 136 yet ELO-favoured over A ranked 40 = high-ELO / modest-rank
divergence -> ELO farmed on qualifier-level fields. Jacquet's entire recent
book is qualifying draws vs 1500-1834 opponents (FO/Wimbledon/Montreal
qualifying). Discount the ELO edge; framework -> lean PASS unless live form
independently backs it.
RECENT FORM (last 5)
Diallo (A, 24, #40): 07-28 Los Cabos hard LOST Boyer(1622) 5-7 2-6 (bad,
straight sets, rusty first hard match off grass); 07-02 Wim grass LOST
Sonego 5set; 06-30 Wim WON Bonzi(1722) ret; 06-25 East LOST Draper(1848);
06-24 East WON Etcheverry(1567). Legit top-50 grass player; thin/poor on
hard. Shape STABLE-to-FADING onto hard (evidence: 5-7 2-6 loss).
Jacquet (B, 25, #136): 08-01 Montreal(quali) FINAL WON Shevchenko(1697) 3set;
07-22 Estoril clay LOST Blockx(1834); 07-21 clay WON Merida(1796); 07-19
Estoril FINAL LOST; 07-18 WON DaRosa(1500). Match-sharp on hard (just
qualified this week) but streak built on weak/qualifier fields. Grindy,
tiebreak-heavy profile.
PATH: A = first Montreal match, 5d rest, off a bad hard loss (mild rust).
B = 3-4 matches deep at venue this week ON this surface = match-sharp, the
models' genuine edge. But opposition was qualifier-level.
FATIGUE (Step 2): B won a 3-SET final YESTERDAY (08-01); 4 matches / 12 sets /
633 min in 14d, back_to_back=1 -> extreme spot. Calibrated 3-set-hangover
haircut -4.5pp (age 25 = baseline, no modulation). B 0.573 -> 0.528.
A rested, no penalty.
MARKET CHOICE
ML B @2.10 (eff 2.045, implied 0.489): pre-fatigue edge 8.4%; POST-fatigue
edge 3.9% -> below 5% bar. And that residual rests on farmed ELO I discount
further toward market -> edge vanishes.
ML A: models say A only 42-43% vs market 53.8% -> A is overvalued by market,
no edge to back A.
Totals: model under-22.5 0.612 / Betfair line 23 under 1.92. But same sim
that overrates B drives this, and B's actual match pattern is tiebreak/3-set
heavy (long, high-total matches) -> under lean contradicted by style; no
trustworthy margin edge. PASS.
Handicap: line -0.5 (pick'em), model handicap null, no independent margin
read. PASS.
AH: not staking a dog >=3.0 (B is 2.10) -> shadow record N/A.
alt_markets: present.
GATES
G1 edge-origin: FAIL(effective) — pre-fatigue B edge 8.4% exists numerically,
but it rests on a farmed-ELO model I discount, and the ACTUAL staking edge
after the mandatory 3-set-hangover haircut is 3.9%, under the 5% bar.
G2 consensus: PASS — 3 models align on B (stake-tier only; full ladder), but
tier is moot with no qualifying edge.
G3 fatal-risk: n/a for ML (never blocks); no bet.
VERDICT: PASS. The only numeric edge is on Jacquet and it comes from an ELO
inflated on qualifier fields (rank 136 "favourite" over rank 40). His genuine
positive (hard-court match-sharpness) is offset by an extreme fatigue spot
(3-set final yesterday, 633 min / 4 matches), which drops the edge below the
bar. The exchange sees all of this and prices Diallo the favourite; I have no
information it is slow to price. No edge -> pass.