TennisEdge's AI model finds no betting value in A. Tabilo vs B. Shelton (Washington on 2026-08-01) — the market is efficient, with B. Shelton favoured at 1.32.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | A. Tabilo | B. Shelton |
|---|---|---|
| Elo ratings-based, surface-weighted |
29.7% | 70.3% |
| Machine learning 29-feature gradient-boosted model |
28.5% | 71.5% |
| Serve / return point-by-point simulation |
27.8% | 72.2% |
| Blend what the bot actually prices off |
27.8% | 72.2% |
| Market implied de-vigged from the opening odds |
26.0% | 74.0% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| A. Tabilo | clay | 1825 | 125–68 | 193 |
| B. Shelton | hard | 1936 | 144–68 | 212 |
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.
A. Tabilo leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 1, 2026 | Washington | hard | A. Tabilo | 2 - 1 |
| L | R. Jodar | 2 - 1 | Aug. 2, 2026 |
| W | B. Shelton | 2 - 1 | Aug. 1, 2026 |
| W | T. Atmane | 0 - 2 | July 30, 2026 |
| W | T. Griekspoor | 2 - 1 | July 27, 2026 |
| L | T. Torres | 0 - 2 | July 22, 2026 |
| L | A. Tabilo | 2 - 1 | Aug. 1, 2026 |
| W | U. Humbert | 1 - 2 | July 30, 2026 |
| W | M. Damm | 1 - 2 | July 28, 2026 |
| L | O. Virtanen | 3 - 2 | June 30, 2026 |
| W | C. Ruud | 0 - 2 | June 25, 2026 |
All Aug. 1, 2026 results & picks →
═══ A. Tabilo (A) vs B. Shelton (B) — Washington QF, hard ═══
MODELS vs MARKET (A = Tabilo)
ELO 0.297 | ML 0.285 | sim 0.278 | exchange-implied 0.277 (A @3.75 → eff 3.61)
→ full convergence on Shelton ~72% / Tabilo ~28%. Market is efficient on the winner.
Ranking cross-check: Shelton #9 vs Tabilo #35; ELO agrees (Shelton ~0.72). Credible.
RECENT FORM (last 5)
Tabilo: W vs Atmane (1755, 2-set) | W vs Griekspoor (1814, 3-set grind) | L Torres (clay) |
L Rublev (clay 3-set) | W Tirante (clay). Transitioned clay→hard for Washington; 2 wins
here, playing decently but grinding. SHARPENING mildly (won last 2 on hard vs Strong ELO).
Shelton: W vs Humbert (1841, 3-set) | W vs Damm (1734, 3-set) | L Virtanen (grass 5-set) |
W Ruud (exo) | L Fritz (grass 3-TB). Grass-title winner (Stuttgart, beat Fritz). Class of
field but both Washington wins went 3 sets vs lower opponents — STABLE, not dominant.
PATH & SHARPNESS
Both match-sharp at venue (2 matches each). Tabilo off clay (TRANSITION) but has 2 hard wins
banked, so timing grooved. No rust risk either side.
FATIGUE (context only — cumulative load is not an edge)
Tabilo: 645 min / 5 matches in 14d, heavy; won a 2-SET match yesterday (~neutral, no haircut).
Shelton: 299 min / 2 matches; won a 3-SET match yesterday → -4.5pp hangover haircut (age 23,
baseline, no age softening). This pushes Shelton's win prob DOWN, i.e. toward Tabilo — but
G1 forbids fatigue creating the staked edge, and the pre-fatigue A edge is already ~0-2%.
NARRATIVE
No first-Final let-down. Shelton TITLE Stuttgart (grass) ~6wk ago — historical, off-surface.
Tabilo clay→hard transition noted (RISK for any tight cover of Tabilo; margins wider).
MARKET CHOICE
ML: A edge ~0-2% (eff 3.61, min 3.78), B negative (0.722 vs implied 0.761). Fail.
Handicap: B -2.5 covers 0.604 vs implied 0.587 = ~1.5%, high-variance tier (needs 10%). Fail.
A +2.5 covers 0.396 vs implied 0.463 = negative. Fail.
Totals: Under 24 model 0.604 vs implied ~0.523 = ~7% nominal — BUT driven entirely by a 3.0-
game bias correction (raw sim 24.9 games → 50/50); matchup-specific scoreline evidence
contradicts it: both players ran 30-32 games in each Washington match this week, and Shelton
is tiebreak-prone (3 TBs Stuttgart SF, 3 TBs Halle QF). Under thesis on two grinders who just
went long is self-negating → G3 veto (totals). Fail.
AH: B -2.5 @ 1.74 (favourite lay, high-variance) — pass.
GATES
G1 edge-origin: FAIL — pre-fatigue ML edge ~0-2% (below 5%); only fatigue would lift it.
G2 consensus: PASS (stake-tier only) — but no edge to stake regardless.
G3 fatal-risk: FAIL (totals) — under thesis negated by tiebreak-heavy, 30+game matchup.
VERDICT: PASS. Full model-market consensus on the winner; ML/handicap/totals all below bar or
self-negating. No articulable, market-slow information to price. Professional pass.