TennisEdge's AI model finds no betting value in K. Boulter vs K. Cross (Toronto on 2026-08-03) — the market is efficient, with K. Boulter favoured at 1.25.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | K. Boulter | K. Cross |
|---|---|---|
| Elo ratings-based, surface-weighted |
69.7% | 30.3% |
| Machine learning 29-feature gradient-boosted model |
76.5% | 23.5% |
| Serve / return point-by-point simulation |
49.9% | 50.1% |
| Blend what the bot actually prices off |
69.7% | 30.3% |
| Market implied de-vigged from the opening odds |
73.2% | 26.8% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
| Player | Surface | Elo | Record | Matches |
|---|---|---|---|---|
| K. Boulter | hard | 1758 | 102–70 | 172 |
| K. Cross | hard | 1657 | 90–66 | 156 |
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.
K. Cross leads 1–0 over 1 meeting. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 3, 2026 | Toronto | hard | K. Cross | 2 - 0 |
| L | K. Cross | 0 - 2 | Aug. 3, 2026 |
| L | A. Krueger | 2 - 1 | July 28, 2026 |
| L | T. C. Grant | 0 - 2 | June 30, 2026 |
| L | L. Fernandez | 1 - 2 | June 21, 2026 |
| L | D. Vekic | 0 - 2 | June 13, 2026 |
| W | K. Boulter | 0 - 2 | Aug. 3, 2026 |
| L | H. Dart | 2 - 0 | July 29, 2026 |
| L | E. Kalieva | 0 - 2 | July 23, 2026 |
| W | W. Osuigwe | 0 - 2 | July 22, 2026 |
| W | S. Saito | 2 - 0 | July 19, 2026 |
All Aug. 3, 2026 results & picks →
K. Boulter (A, rank 60, age 30) vs K. Cross (B, rank 199, age 21) — WTA Toronto R1 (1/64), hard.
MODELS vs MARKET
Betfair ML: A 1.35 (implied 74.1%), B 3.75 (26.7%). Book: A 1.25, B 3.42. Pinnacle no-vig A 71.8%.
ELO 69.7% A | ML 76.5% A | model 69.7% A. Value_pick B edge 3.5% is vs soft BOOK 3.42;
vs commission-adj Betfair 3.75 (eff 3.61) B edge = 30.3% - 27.7% = 2.6%, below the 5% bar.
Resimulate baseline win_prob_a = 49.9% — a huge divergence from ELO/ML/market/ranking. Cause:
contaminated sim inputs. Cross spw .567/rpw .432 were FARMED at W75 Granby (opp ELO 1308-1623);
Boulter spw .589/rpw .41 built vs elite grass/clay (Rybakina 1869, Potapova 1918, Vekic 1767).
The sim materially OVER-rates Cross → its handicap/totals probs are biased toward B and untrustworthy.
RECENT FORM / SHAPE (top 5)
Boulter (A): FADING but tour-class. L Krueger (1708) 3-6 6-2 6-2 (Washington, hard, 07-28);
L Grant (Wim); L Fernandez (Bad Homburg); L Vekic 1-6 3-6 (London SF); W Rybakina 1869 (London QF).
One strong run (London SF, beat Rybakina) but poor since; only recent hard match was a 3-set loss.
Cross (B): level-dropper, FADING. L Dart 1687 6-2 6-1 (Vancouver hard, 07-29 — blowout);
L Kalieva 1500 3-6 2-6 (Evansville — blowout); WON W75 Granby title 07-19 (weak field 1300-1620);
vs QUALITY opponents she gets run over: 6-2 6-1 (Dart), 3-6 2-6 (Kalieva), 6-4 6-0 (Sramkova FO),
3-6 4-6 (Stoiana Wim). Her ELO/serve numbers are inflated by the W75 field.
ELO/RANK SANITY: Cross ELO/sim-competitiveness is farmed at W75 — discount. Boulter rank 60 vs 199,
market and ML correctly favour her ~74-76%. No mispricing to exploit against the favourite.
FATIGUE: Cross matches_14d 5, court 386min, back_to_back 2 — heavy load BUT last match 07-29 (4 days
rest), no extreme spot; age 21 recovers fast → context only, not an edge. Boulter last 07-28, fine.
No 3-set-hangover haircut applies to either side.
MARKET CHOICE
ML A @1.35: my win prob ~0.74 → min_odds 1.05/0.74 = 1.42 > 1.35 → auto-PASS (below min).
ML B @3.75: eff edge 2.6% < 5%, and Cross is over-rated even at 30% → PASS.
Handicap A -4.5 @1.99: high-variance (|line|>=4.5, bar 10%). Sim (biased) says A covers only ~30% at -3.5;
a fading, scrappy favourite laying -4.5 is unsupported → PASS.
Handicap B +4.5 @1.92: sim B-covers ~65% is INFLATED by farmed inputs; margin evidence CONTRADICTS it —
Cross's losses to quality opponents are blowouts (5-9 game margins). Cover thesis self-negates → G3 VETO.
Totals O 20.5 @1.97: sim over 53.7% biased; eff edge ~1.7% < 5% → PASS.
AH: B +4.5 @ 1.92 (evaluated, vetoed).
alt_markets: present (handicap/totals/betfair_ml all live).
GATES (best candidate = B +4.5 handicap)
G1 edge-origin: PASS — nominal numeric edge exists from the sim (pre-narrative), but the sim inputs
are contaminated (49.9% vs 74% consensus), so the edge is not real.
G2 consensus: PASS(tier) — lone contaminated sim signal → would cap 0.5-1.0u even if it survived.
G3 fatal-risk: FAIL — margin bet; Cross's blowout-loss distribution vs quality opponents self-negates
the +4.5 cover. Hard veto for handicap/totals → PASS.
VERDICT: PASS. Market efficiently prices Boulter ~74%; the only nominal "value" (Cross ML / Cross +4.5)
is an artifact of a farmed-W75 ELO and contaminated sim serve/return inputs. No market clears its bar
with trustworthy numbers. No bet.