TennisEdge's AI model finds no betting value in M. Trungelliti vs N. Mejia (Montreal on 2026-08-01) — the market is efficient, with M. Trungelliti favoured at 1.40.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | M. Trungelliti | N. Mejia |
|---|---|---|
| Elo ratings-based, surface-weighted |
43.6% | 56.4% |
| Machine learning 29-feature gradient-boosted model |
50.5% | 49.5% |
| Serve / return point-by-point simulation |
27.9% | 72.1% |
| Blend what the bot actually prices off |
48.2% | 51.8% |
| Market implied de-vigged from the opening odds |
66.3% | 33.7% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| M. Trungelliti | clay | 1739 | 160–87 | 247 |
| N. Mejia | hard | 1591 | 88–100 | 188 |
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 22, 2026.
N. Mejia leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 1, 2026 | Montreal | hard | N. Mejia | 2 - 1 |
| L | N. Mejia | 1 - 2 | Aug. 1, 2026 |
| L | Y. Hanfmann | 0 - 2 | July 22, 2026 |
| W | J. M. Cerundolo | 0 - 2 | July 21, 2026 |
| L | A. Davidovich Fokina | 0 - 2 | July 15, 2026 |
| W | K. Jacquet | 0 - 2 | July 13, 2026 |
| W | M. Landaluce | 2 - 1 | Aug. 3, 2026 |
| W | M. Trungelliti | 1 - 2 | Aug. 1, 2026 |
| L | A. Hernandez | 0 - 2 | July 26, 2026 |
| W | M. Soto | 2 - 0 | July 12, 2026 |
| W | J. P. Varillas | 2 - 1 | July 11, 2026 |
All Aug. 1, 2026 results & picks →
═══ Trungelliti vs Mejia — Montreal (hard), Final ═══
MODELS vs MARKET
Market: Trungelliti @1.40 (66.3%) / Mejia @2.75 (36.4%)
ELO A 0.436 | ML A 0.505 | blended A 0.482
All three models put Trungelliti at/below coinflip → they favor Mejia vs a
market that has Trungelliti a strong 66% favourite. value_pick B, edge ~15-18%.
betfair_ml: null — NO exchange liquidity on this event.
RECENT FORM (last 5)
Trungelliti (A, 36, rank 78): all recent play CLAY/GRASS — Kitzbuhel (clay,
last match 07-22, ~10d ago), Umag (clay), Wimbledon (grass), Eastbourne (grass).
Montreal is HARD → TRANSITION + rust (>7d idle). No recent hard-court results.
Mejia (B, 26, rank 168): Los Cabos SF (hard) 07-26 LOST 5-7 2-6 to a 1377-elo
opponent — poor hard-court shape; before that won Bogota (CLAY Challenger) title
vs 1322-1581 fields. His rating is built on clay/weak fields, not hard.
FATIGUE
A: last match 07-22, matches_7d 0 — rested, no extreme spot (rust, not fatigue).
B: last match 07-26, 1 match/7d — rested. No 3-set hangover on either side.
Neither side qualifies for a fatigue edge.
ELO/RANKING SANITY (decisive)
Trungelliti: rank 78 yet ELO makes him a dog to rank-168 Mejia → "strong rank,
low ELO" (off-surface, layoff) → ELO UNDERSTATES him; do NOT back opponent on gap.
Mejia: rank 168 yet high ELO → "high ELO, modest rank" → farmed at a weak clay
Challenger; discount unless live form backs it — and his last hard match was a
bad loss to a 1377 player, so live form does NOT back the ELO edge.
Both checks push the model-B edge toward discount/PASS.
GATES
G1 edge-origin: FAIL — the numeric model-vs-market edge on B is driven by farmed
ELO (weak-field clay Challenger) that the ELO/ranking sanity + live hard-court
form both invalidate; no trustworthy pre-narrative edge survives.
G2 consensus: stake-tier only — would be lone/contaminated (farmed ELO, off-surface).
G3 fatal-risk: n/a (ml candidate).
PRICING
No betfair_ml — illiquid small event = the segment the backtest loses on
(Pricing Discipline #4 → prefer PASS). Book B edge exists nominally (~15% vs 2.75)
but rests on a discountable rating, not confirmable value.
VERDICT: PASS. Large model-market gap, but the B edge is manufactured by farmed/
understated ELO ratings on an off-surface transition spot with no exchange price
to confirm. Market favoring the higher-ranked, if rusty, Trungelliti is defensible;
no trustworthy edge to stake.