TennisEdge's AI model finds no betting value in T. Fritz vs A. Michelsen (Washington on 2026-07-31) — the market is efficient, with T. Fritz favoured at 1.58.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | T. Fritz | A. Michelsen |
|---|---|---|
| Elo ratings-based, surface-weighted |
55.7% | 44.3% |
| Machine learning 29-feature gradient-boosted model |
48.6% | 51.4% |
| Serve / return point-by-point simulation |
61.7% | 38.3% |
| Blend what the bot actually prices off |
56.7% | 43.3% |
| Market implied de-vigged from the opening odds |
60.0% | 40.0% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| T. Fritz | hard | 1972 | 179–93 | 272 |
| A. Michelsen | hard | 1895 | 148–85 | 233 |
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. 3, 2026.
A. Michelsen leads 2–1 over 3 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| July 31, 2026 | Washington | hard | T. Fritz | 2 - 1 |
| March 9, 2026 | ATP Indian Wells | hard | A. Michelsen | 2 - 0 |
| May 22, 2024 | ATP Geneva | clay | A. Michelsen | 2 - 1 |
| W | R. Jodar | 2 - 0 | Aug. 3, 2026 |
| W | B. Nakashima | 1 - 2 | Aug. 1, 2026 |
| W | A. Michelsen | 2 - 1 | July 31, 2026 |
| W | K. Majchrzak | 2 - 0 | July 29, 2026 |
| W | Z. Bergs | 2 - 0 | July 27, 2026 |
| W | J-L. Struff | 2 - 1 | Aug. 4, 2026 |
| L | T. Fritz | 2 - 1 | July 31, 2026 |
| W | A. Mannarino | 2 - 0 | July 29, 2026 |
| W | M. McDonald | 2 - 0 | July 28, 2026 |
| W | J. Fearnley | 2 - 0 | July 26, 2026 |
All July 31, 2026 results & picks →
T. Fritz (A, 28, rank 9) vs A. Michelsen (B, 21, rank 42) - Washington, hard.
MODELS
ELO: Fritz 55.7pct | ML: Fritz 48.6pct (lone outlier, favours Michelsen) | Ensemble: Fritz 56.7pct
Sim (resimulate): Fritz 61.7pct (spw .616/.608, matchup-adj); expected margin +1.6 games.
Market: book 1.58/2.37 (Fritz ~60pct de-vig); Betfair ML odds_a 1.03 BROKEN/stale, odds_b 2.5 (~40pct).
Value_pick B (Michelsen) edge 3.3pct rests SOLELY on the break-even ML outlier; ELO, ensemble, sim
and market all cluster Fritz 56-62pct. Consensus is Fritz, not the value pick.
RECENT FORM (last 5)
Fritz: W Majchrzak(1763) 6-3 6-4 DOM, W Bergs(1766) 6-3 6-4 DOM, L Zverev(1913) grass,
W Bublik(1856) grass, W Sonego(1675) grass. Two straight-set DOM over Strong opponents at THIS
event -> SHARPENING on hard.
Michelsen: W Mannarino(1718) 6-4 7-6 GRIND, W McDonald(1695) 6-4 7-5, W Fearnley(1702) 7-6 6-3
(Bloomfield Hills TITLE 07-26), W Suresh(1666) GRIND, W Lajal(1705) DOM. In form but over
Avg/lower-Strong fields incl. a Challenger. Mixed DOM/GRIND -> STABLE, weaker field.
PATH / SHARPNESS
Fritz: 2 matches deep at Washington, both DOM. Match-sharp, well-rested, US hard specialist.
Michelsen: 7 matches in 8 days (won Bloomfield Hills Challenger title then 2 at Washington),
back_to_back=5, 774 court-min/14d. Today is his 3rd match in 3 consecutive days (07-28/29/30),
an extreme fatigue spot. BUT age 21 (<=22) recovers fast, softens one tier to slight-edge Fritz.
Yesterday was a 2-set win (no 3-set hangover haircut). Fatigue is CONTEXT here, leans Fritz, not
a bettable edge on its own.
NARRATIVE
No first-Final let-down (Michelsen title was a Challenger; backed up with 2 wins). Fritz off grass
swing but adjusted (2 hard wins). No transition concern. ELO approx rank both (Fritz 9/55.7pct,
Michelsen 42/44pct) -> ELO credible.
GATES
G1 edge-origin: FAIL - only numeric edge (value_pick B 3.3pct) comes solely from the lone ML
outlier and is below the 5pct bar; ELO/ensemble/sim/market all price Fritz 56-62pct.
G2 consensus: stake-tier only - lone-ML support, contaminated by convergence against it.
G3 fatal-risk: n/a (ML market; no margin bet).
VERDICT: PASS. Four of five estimates (ELO, ensemble, sim, market) converge on Fritz ~56-62pct; the
lone dissent (ML, break-even) points at Michelsen but is contradicted by the sim AND the qualitative
read (Fritz sharper vs stronger fields, well-rested; Michelsen on 3rd match in 3 days after a title
run). No value on Michelsen at 2.5 (edge ~2-3pct off credible models), no executable/value price to
back Fritz (book rich, Betfair 1.03 broken). Efficient market, no edge.