TennisEdge's AI model finds no betting value in C. Ugo Carabelli vs C. Norrie (Montreal on 2026-08-03) — the market is efficient, with C. Norrie favoured at 1.16.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | C. Ugo Carabelli | C. Norrie |
|---|---|---|
| Elo ratings-based, surface-weighted |
23.3% | 76.7% |
| Machine learning 29-feature gradient-boosted model |
27.7% | 72.3% |
| Serve / return point-by-point simulation |
40.0% | 60.0% |
| Blend what the bot actually prices off |
27.0% | 73.0% |
| Market implied de-vigged from the opening odds |
18.4% | 81.6% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| C. Ugo Carabelli | clay | 1796 | 208–111 | 319 |
| C. Norrie | hard | 1843 | 125–89 | 214 |
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 20, 2026.
C. Norrie leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 4, 2026 | Montreal | hard | C. Norrie | 2 - 0 |
| L | C. Norrie | 0 - 2 | Aug. 4, 2026 |
| L | T. Droguet | 2 - 1 | July 20, 2026 |
| L | R. A. Burruchaga | 2 - 1 | July 16, 2026 |
| W | P. Carreno-Busta | 1 - 2 | July 15, 2026 |
| W | M. Topo | 1 - 2 | July 14, 2026 |
| W | C. Ugo Carabelli | 0 - 2 | Aug. 4, 2026 |
| L | D. Shapovalov | 1 - 2 | Aug. 1, 2026 |
| W | B. Tomic | 2 - 0 | July 31, 2026 |
| W | J. Duckworth | 2 - 0 | July 30, 2026 |
| W | A. Kovacevic | 2 - 0 | July 28, 2026 |
All Aug. 3, 2026 results & picks →
═══ Ugo Carabelli (A) vs Norrie (B) — Montreal R1 (hard) ═══
MODELS vs MARKET
ELO 23.3% | ML 27.7% | blended 27.0% for A.
Book: A 5.15 / B 1.16. Betfair exchange: A 4.2 / B 1.15. Pinnacle no-vig A 22.14%.
Book 5.15 flags A as "value", but the sharp Betfair price (4.2 -> implied 23.8%,
commission-adjusted 24.75%) has already corrected the soft book. Edge vs executable
Betfair = 27.7% - 24.75% = ~3.0% (blended 27% -> ~2.25%). Below the 5% bar.
RECENT FORM
Ugo Carabelli (A, 27, rk 61) — CLAY SPECIALIST. Last 5:
L Droguet (clay 1792); L Burruchaga (clay 1798); W Carreno-Busta (clay 1850, GRIND);
W Topo (clay 1544, GRIND tiebreaks); L Merida (grass). All wins on clay
(Berrettini/Tiafoe/CarrenoBusta earlier). TRANSITION clay->hard + 13 days off (rust).
ELO is clay-farmed; hard-court level unproven and likely lower than models imply.
Norrie (B, 30, rk 28) — MATCH-SHARP on hard. Last 5:
L Shapovalov (hard 1829, SF yesterday, 3 sets); W Tomic 6-1 6-0 DOM; W Duckworth 6-4 6-4;
W Kovacevic 6-4 6-1; UTS exhibition. Reached Los Cabos SF on hard — genuine hard-courter.
FATIGUE
Norrie: heavy load (6 matches/14d, 1155 court min, back_to_back 2), age 30 (recovers slowly).
BUT he LOST yesterday's 3-setter -> no -4.5pp win-hangover applies. Cumulative load alone is
context, not an edge; and the market over-prices visible fatigue -> value, if any, is ON the
tired side (Norrie), not against. Carabelli well-rested but rusty (last match 07-20).
NARRATIVE
Surface transition A (clay->hard) = RISK on the side we'd back, not support.
Norrie SF loss yesterday, no let-down peak (he lost). No title recency.
GATES
G1 edge-origin: FAIL — edge vs executable Betfair (~2-3%) is below the 5% bar BEFORE any
narrative; the only thing pushing toward A is Norrie-fatigue, which cannot create edge.
G2 consensus: n/a (G1 already fails) — models cluster near the exchange price anyway.
G3 fatal-risk: n/a (ML) — but self-negating: backing a rusty clay-courter on hard vs a
match-sharp hard-courter; the fatigue thesis is the market's, and it's over-priced.
VERDICT: PASS. Book 5.15 looks soft but Betfair 4.2 already prices A fairly; ~3% edge is below
the 5% bar (G1 fail). Qualitatively the clay->hard transition + rust argues A is over-, not
under-rated. No bettable edge.