TennisEdge's AI model finds no betting value in A. Vukic vs T. Barrios Vera (Montreal on 2026-08-01) — the market is efficient, with A. Vukic favoured at 1.66.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | A. Vukic | T. Barrios Vera |
|---|---|---|
| Elo ratings-based, surface-weighted |
65.8% | 34.2% |
| Machine learning 29-feature gradient-boosted model |
62.7% | 37.3% |
| Serve / return point-by-point simulation |
62.0% | 38.0% |
| Blend what the bot actually prices off |
61.1% | 38.9% |
| Market implied de-vigged from the opening odds |
55.9% | 44.1% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| A. Vukic | hard | 1732 | 146–124 | 270 |
| T. Barrios Vera | clay | 1624 | 161–103 | 264 |
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. 4, 2026.
Level at 1–1 over 2 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 1, 2026 | Montreal | hard | A. Vukic | 2 - 0 |
| March 17, 2026 | Miami | hard | T. Barrios Vera | 2 - 0 |
| L | D. Altmaier | 1 - 2 | Aug. 4, 2026 |
| W | T. Barrios Vera | 2 - 0 | Aug. 1, 2026 |
| L | L. Musetti | 1 - 2 | July 29, 2026 |
| W | Z. Svajda | 2 - 0 | July 27, 2026 |
| W | A. Bolt | 2 - 0 | July 26, 2026 |
| L | A. Vukic | 2 - 0 | Aug. 1, 2026 |
| L | D. Dietrich | 2 - 1 | July 21, 2026 |
| L | N. Visker | 0 - 2 | July 14, 2026 |
| L | I. Simakin | 1 - 2 | July 7, 2026 |
| L | D. Sweeny | 3 - 2 | June 25, 2026 |
All Aug. 1, 2026 results & picks →
═══ A. Vukic (A) vs T. Barrios Vera (B) — Montreal, HARD, Final ═══
MODELS vs MARKET
ELO 65.8% | ML 62.7% | blended model 61.1% — all favour A
Market: odds_a 1.66 (raw implied 60.2%), devigged implied_a 55.9%; value_pick A, value_edge 5.3%.
All three models CONSENSUS-agree and sit ~5pp above the devigged market. Modest, not large.
RECENT FORM (last 5 shown; classification uses full window)
A. Vukic (30, rk 99) — RED-HOT on hard:
07-29 L Musetti (1916, Elite) 3-6 7-5 3-6 — competitive loss to elite
07-27 W Svajda (1755, Strong) 6-3 6-1 DOM
07-26 W Bolt (1641, Avg) 6-0 6-1 DOM
07-25 W Dougaz (1606, Avg) 6-3 6-3 DOM
07-19 W Gea (1813, Strong) 6-4 1-6 6-1 — Granby title run
Shape: SHARPENING, match-sharp (deep hard-court runs, two recent titles Granby/Washington).
T. Barrios Vera (28, rk 134) — FREEFALL + rusty:
07-21 L Dietrich (1601) 6-4 4-6 6-1
07-14 L Visker (1357) 4-6 4-6
07-07 L Simakin (1202) 6-4 5-7 6-7
06-25 L Sweeny (1726) Wimbledon Final (5 sets)
06-24 W Jubb (1571)
Shape: FADING hard — last 3 are losses to sub-1600/1357/1202 ELO opponents (clay). 10 days idle (last 07-21) → rust risk. Recent hard-court form absent.
FATIGUE
A: 7 matches/14d, 727 court-min, back_to_back 3, last 07-29 (a LOST 3-setter — no 3-set-WIN hangover haircut applies). Age 30 (≥30, slow recovery). Heavy cumulative load, BUT per calibration cumulative minutes alone = noise/context, not an edge; market tends to over-price this visible-fatigue story. A had rest before the final.
B: 1 match/14d, idle since 07-21 — rusty, not tired.
Net: fatigue is context only; does not create an edge either way.
NARRATIVE
Surface: match on HARD; B's recent matches all CLAY → TRANSITION for B (level-depressing, supports A directionally on ML).
No let-down spot (A's last was a loss; B's Wimbledon final was 06-25, >5w ago).
Venue: Montreal — no prior data in window.
ELO/ranking: A rk 99 ELO-fav — coherent, no farmed-rating flag. B rk 134 but ELO from May peaks; recent results far below that → ELO likely OVERSTATES B, i.e. B may be even weaker than priced.
GATES
G1 edge-origin: [PASS] — ~5pp model-vs-market gap exists pre-narrative (ELO edge 5.6%), origin is model disagreement, not fatigue.
G2 consensus: [PASS] — all 3 models agree; full-tier eligible (stake-tier gate only).
G3 fatal-risk: [PASS] — N/A, ML market.
PRICING (the blocker)
alt_markets: none at analysis time — no executable Betfair price.
Rule #4 (no exchange): ML needs edge >= 8% vs book quote, else prefer PASS/LEAN.
Edge vs raw book 1.66: model 0.9%, ML 2.5%, ELO 5.6% — best case ~5.6%, BELOW the 8% book-only bar.
Cannot BET at the book price with no exchange to confirm/execute.
MARKET CHOICE
ml only (alt_markets null). AH: not quoted.
alt_markets: none at analysis time
refutation_check: no web search tool available in environment and brief has no web_insights/player_notes — check could not be run; lean not upgraded.
VERDICT: LEAN A. Vukic.
Genuine, articulable qualitative edge — A is a twice-recent hard-court finalist and SHARPENING; B is in freefall (three straight losses to 1601/1357/1202 ELO on clay) and comes in rusty and transitioning off clay. All three models confirm A above market. The ONLY thing stopping a bet is the missing exchange price + the sub-8% book-only edge. Flag for betday re-check when Betfair develops: if A holds ~1.66+ on the exchange, edge becomes executable.