TennisEdge's AI model finds no betting value in M. Berrettini vs M. Navone (Montreal on 2026-08-03) — the market is efficient, with M. Berrettini favoured at 1.44.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | M. Berrettini | M. Navone |
|---|---|---|
| Elo ratings-based, surface-weighted |
74.2% | 25.7% |
| Machine learning 29-feature gradient-boosted model |
51.9% | 48.1% |
| Serve / return point-by-point simulation |
64.5% | 35.5% |
| Blend what the bot actually prices off |
74.2% | 25.7% |
| Market implied de-vigged from the opening odds |
65.6% | 34.4% |
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. Berrettini | hard | 1804 | 70–53 | 123 |
| M. Navone | clay | 1815 | 195–108 | 303 |
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. Berrettini leads 3–1 over 4 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 3, 2026 | Montreal | hard | M. Navone | 2 - 1 |
| April 30, 2026 | Cagliari | clay | M. Berrettini | 2 - 0 |
| April 7, 2025 | Monte Carlo | clay | M. Berrettini | 2 - 0 |
| April 6, 2024 | ATP Marrakech | clay | M. Berrettini | 2 - 1 |
| L | M. Navone | 1 - 2 | Aug. 3, 2026 |
| L | G. Dimitrov | 3 - 2 | July 4, 2026 |
| W | A. Fils | 3 - 1 | July 2, 2026 |
| W | S. Wawrinka | 1 - 3 | June 30, 2026 |
| L | M. Arnaldi | 0 - 1 | June 3, 2026 |
| W | M. Berrettini | 1 - 2 | Aug. 3, 2026 |
| L | Q. Halys | 0 - 2 | July 23, 2026 |
| W | J-L. Struff | 0 - 2 | July 22, 2026 |
| W | A. Muller | 0 - 2 | July 20, 2026 |
| L | S. Travaglia | 2 - 0 | July 15, 2026 |
All Aug. 3, 2026 results & picks →
=== M. Berrettini (A) vs M. Navone (B) - Montreal, HARD, 1/64-finals ===
MODELS vs MARKET
ELO A 74.25pct | ML A 51.95pct | model A 74.25pct
Book: A 1.44 / B 2.75 | Betfair (executable): A 1.60 / B 2.62
Betfair devig ~ A 62.5pct / B 38pct. Book drifted 1.44 -> 1.60 on exchange = sharps
already discounting Berrettini toward the rust/transition story.
Huge model split: ELO 74pct vs ML 52pct (22pp). Market sits between, nearer ELO.
ELO/RANK SANITY
Berrettini ELO-fav (74pct) yet ranked 100; Navone ranked 44. NOT farmed ELO -
Berrettini's rank is depressed by injury layoffs; his ELO is built on real wins
vs Strong/Elite (Fils 1721, Cerundolo 1847). ELO credible, arguably understates.
Navone rank 44 is clay-inflated (Argentine clay grinder), weak on hard.
RECENT FORM (last 5 each)
Berrettini (all grass/clay - none on hard):
07-04 Wimb R16 L Dimitrov 1799 6-3 6-4 3-6 5-7 6-3 (5s, close vs Strong)
07-02 Wimb R32 W Fils 1721 GRIND (4s) over Strong
06-30 Wimb R64 W Wawrinka 1511 GRIND (all TBs) over Avg - under-performing
06-03 FO QF L Arnaldi 1865
06-01 FO R16 W Cerundolo 1847 GRIND over Elite
Navone (all clay except Wimbledon):
07-23 Kitz QF L Halys 1773 5-7 3-6
07-22 Kitz R16 W Struff 1703 DOM (clay)
07-20 Kitz R32 W Muller 1643 DOM (clay)
07-15 Bastad R16 L Travaglia 1612
07-01 Wimb R64 L Cobolli 1640 (grass)
SHAPE / PATH
Berrettini: FADING-to-neutral - grindy grass wins, then 5-set loss; now ~29 days
off. RUST RISK (match one after a month) + grass->hard TRANSITION => depressed
level + wider margin variance. First hard match of the swing.
Navone: recent clay form OK (two DOM at Kitzbuhel) but that form is CLAY and does
NOT transfer to hard; clay->hard TRANSITION. Last match 10d ago, no live fatigue.
Both players transitioning; neither has a hard-court result in the window.
FATIGUE
Neither in an extreme spot. Berrettini fully rested (rust, not fatigue). Navone
last played 07-23 (10d), well recovered. No fatigue edge either way.
NARRATIVE
No let-down. Surface transition on BOTH sides (rust angle already in the drifted
exchange price). Venue: neither has Montreal momentum in window.
ALT MARKETS (Betfair)
Handicap A -2.5 @1.94 (model A covers 54.7pct): edge ~ +1.9pct, below bar AND
high-variance lay tier (needs >=10pct). PASS.
Total 22.5, Under @1.94 (model under 52.9pct): edge ~ +0.1pct. PASS.
ML A @1.60 eff 1.57 (breakeven 63.7pct): ELO gives ~+10.5pct but ML says NEGATIVE.
AH: A -2.5 @1.94 (not staked)
GATES (ML A candidate)
G1 edge-origin: BORDERLINE - raw ELO 74pct vs market 62.5pct is a real
model-market gap, but hinges entirely on ELO being right while ML (52pct)
disagrees by 22pp.
G2 consensus: lone-model (ELO only); ML flatly opposes -> stake tier capped.
G3 fatal-risk: ML bet, not blocking; but rust+transition on the rusty favourite
is a genuine thesis-negating risk in match one.
VERDICT: PASS
The staked edge depends on trusting ELO alone against a 22pp ML disagreement, with
the favourite carrying real rust (29d off) + grass->hard transition in his first
match of the swing. The exchange has already drifted 1.44 -> 1.60, pricing that
rust; 1.60 (62.5pct) is a reasonable estimate I cannot confidently beat. Navone at
2.62 isn't value either - clay specialist on hard, also transitioning. No confident,
articulable edge over the executable price. PASS.