TennisEdge's AI model finds no betting value in S. Sierra vs Q. Zheng (Bad Homburg on 2026-06-22) — the market is efficient, with Q. Zheng 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 | S. Sierra | Q. Zheng |
|---|---|---|
| Elo ratings-based, surface-weighted |
52.6% | 47.4% |
| Machine learning 29-feature gradient-boosted model |
30.4% | 69.6% |
| Blend what the bot actually prices off |
41.5% | 58.5% |
| Market implied de-vigged from the opening odds |
32.3% | 67.7% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
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 May 29, 2026.
Q. Zheng leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| June 22, 2026 | Bad Homburg | grass | Q. Zheng | 2 - 1 |
| L | S. Stephens | 0 - 2 | Aug. 4, 2026 |
| L | T. Prozorova | 1 - 2 | July 27, 2026 |
| L | C. Gauff | 1 - 2 | July 1, 2026 |
| W | A. Bondar | 2 - 1 | June 29, 2026 |
| L | Q. Zheng | 1 - 2 | June 22, 2026 |
| L | L. Tararudee | 2 - 0 | Aug. 1, 2026 |
| L | A. Eala | 2 - 1 | July 28, 2026 |
| L | B. Krejcikova | 2 - 0 | July 17, 2026 |
| W | E. Micic | 0 - 2 | July 15, 2026 |
| W | J. Bouzas Maneiro | 2 - 0 | July 13, 2026 |
All June 22, 2026 results & picks →
The model consensus suggests a 5.9% edge on underdog S. Sierra, but this is entirely driven by an ELO model projection (52.6% win chance) that rates the match as a coin flip. This ELO projection is statistically unsound given the players' profiles: S. Sierra holds a Hard Court ELO Rank of 448, while Q. Zheng is ranked 24th. The ML model (30.4% for Sierra) aligns closely with the market (34%), indicating the market has priced this match correctly. Without evidence of injury for Zheng or a specific surface advantage for Sierra, the ELO model's outlier prediction appears to be an error caused by overfitting to Sierra's small sample of 13 grass matches. We override the mathematical value recommendation.