TennisEdge's AI model finds no betting value in A. Popyrin vs T. Kokkinakis (Montreal on 2026-08-01) — the market is efficient, with T. Kokkinakis favoured at 1.53.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | A. Popyrin | T. Kokkinakis |
|---|---|---|
| Elo ratings-based, surface-weighted |
27.3% | 72.7% |
| Machine learning 29-feature gradient-boosted model |
51.5% | 48.5% |
| Serve / return point-by-point simulation |
16.2% | 83.8% |
| Blend what the bot actually prices off |
27.3% | 72.7% |
| Market implied de-vigged from the opening odds |
39.2% | 60.8% |
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. Popyrin | hard | 1665 | 84–96 | 180 |
| T. Kokkinakis | hard | 1789 | 96–56 | 152 |
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. 1, 2026.
T. Kokkinakis leads 3–1 over 4 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 1, 2026 | Montreal | hard | A. Popyrin | 1 - 0 |
| May 28, 2024 | ATP French Open | clay | T. Kokkinakis | 3 - 2 |
| Feb. 17, 2023 | Manama | hard | T. Kokkinakis | 2 - 0 |
| Jan. 9, 2023 | Adelaide 2 | hard | T. Kokkinakis | 2 - 1 |
| W | T. Kokkinakis | 1 - 0 | Aug. 1, 2026 |
| L | A. Andrade | 2 - 1 | July 29, 2026 |
| L | J. Lehecka | 3 - 0 | June 30, 2026 |
| L | J. Choinski | 2 - 1 | June 22, 2026 |
| L | R. Collignon | 0 - 2 | June 16, 2026 |
| L | A. Popyrin | 1 - 0 | Aug. 1, 2026 |
| L | A. Walton | 1 - 2 | July 22, 2026 |
| W | E. Zhu | 0 - 2 | July 21, 2026 |
| L | A. Bublik | 2 - 3 | June 30, 2026 |
| W | A. Rublev | 1 - 2 | June 24, 2026 |
All Aug. 1, 2026 results & picks →
MONTREAL FINAL (hard) — A. Popyrin (A, rank 60) vs T. Kokkinakis (B, rank 454)
MODELS vs MARKET
Odds: A 2.37 (impl 42.2% raw) / B 1.53 (impl 65.4% raw; norm B 60.8%)
ELO_prob_A 27.3% -> B 72.7% (ELO strongly favours B)
ML_prob_A 51.6% -> B 48.5% (ML sees a coin-flip)
model_prob_A (=ELO) 27.3%. value_pick B, value_edge +11.9% off ELO.
The two models DISAGREE by ~24pp on B. That is the whole story.
ELO/RANKING SANITY (decisive)
Kokkinakis is a 72.7% ELO favourite while ranked 454. Classic "high ELO,
modest rank" — ELO farmed at weak events. His path confirms it: recent
slate is Challenger/qualification level (Bloomfield Hills, Zagreb, Bratislava,
Dublin quali) vs opponents 1434-1752. That ELO edge is inflated by weak
fields -> discount it, do not lean on it.
RECENT FORM / SHAPE
Popyrin (A): FADING hard. Last 5 = five straight losses, incl. losses to
WEAK opponents — 07-29 Vancouver L Andrade (1560) 7-5 3-6 6-1; 06-22 L
Choinski (1524) 1-6 6-2 6-2; 05-26 FO L Svajda (1317). Last win was mid-May.
ML's 51.6% for Popyrin is not credible against this collapse (stale
season-long serve inputs).
Kokkinakis (B): mixed/Challenger-level. 07-22 L Walton (1752) 6-2 3-6 6-7;
07-21 W Zhu (1434); 06-30 Wimbledon L Bublik in 5. Zagreb clay title 05-11
(~80d, off-surface, weak field). Path = match play but at low level; last
match 10d ago -> mild rust.
FATIGUE (Step 2)
A: 1 match/14d, 3 sets, 3d rest — fresh. B: last match 07-22 (10d), no
extreme spot either side. Non-factor. No 3-set-hangover applies.
PRICING / MARKETS
alt_markets: none at analysis time (betfair_ml, handicap, totals all null).
No executable exchange price -> ML would be book-priced only (Pricing #4:
needs >=8% book edge AND flag; this segment is the backtest's losing one).
VERDICT
No cleanly-validated edge on either side:
- B bet: edge is from ELO, but ELO is farmed at weak Challenger fields
(rank 454) and ML flatly disagrees -> discounted to ~0.
- A bet: ML edge (51.6% vs 42.2%) is contradicted by ELO AND by Popyrin's
six-loss form collapse -> not credible.
- No exchange liquidity, models split, no qualitative tiebreak favouring
the priced side. Market (B 1.53) is the sharpest estimate here and I have
no specific information it is slow to price.
Decision: PASS.