TENNISEDGE.GG

H. Kaji vs T. Preston — Prediction & Analysis

Vancouver · Vancouver - 1/16-finals · July 26, 2026 · 21:30 UTC

TennisEdge's AI model finds no betting value in H. Kaji vs T. Preston (Vancouver on 2026-07-26) — the market is efficient, with T. Preston favoured at 1.25.

Surface hard Odds H. Kaji 3.75 · T. Preston 1.25

Verdict: no bet

The model found no value at current prices — the market has this one priced about right. That verdict is part of the public record too.
Profiles: H. Kaji · T. Preston · head-to-head record

Full analysis

H. Kaji (A) vs T. Preston (B) — Vancouver, hard, 1/16-finals (WTA)

MODELS
  Market: A 3.75 / B 1.25 (Betfair A 3.90 / B 1.30) — B ~77% fav
  ELO: A 0.251 / B 0.749  |  ML: A 0.503 (break-even split, outlier)  |  model 0.251
  Sim (baseline): A 0.453 / B 0.547, exp games 20.7, margin -0.4 — near coin flip
  KEY DIVERGENCE: sim+ML say ~50/50; ELO+market say B ~75-78%. ELO is opponent-
  adjusted and agrees with the price; the sim/ML are the outliers.

WHY THE SIM IS BIASED (input inspection + re-run)
  Kaji's serve/return (spw 0.574, rpw 0.419) is FARMED ON WEAK ITF FIELDS: her
  last ~15 matches are W35/W50/W75/W100 events vs 1400-1600 ELO opponents.
  Preston's stats (spw 0.566, rpw 0.436) were EARNED AT WTA TOUR LEVEL — Wimbledon
  QF, Nottingham final, beating 1650-1740 ELO opponents. Kaji's numbers are
  inflated, Preston's understated -> the unadjusted sim wrongly reads a coin flip.
  RE-RUN (logged): shaded Kaji spw 0.574->0.554, rpw 0.419->0.399 (weak-field
  correction) to land win prob on the ELO anchor: B 0.744, exp games 19.8,
  margin -2.9. This coherent distribution drives all market reads below.

RECENT FORM (most-recent 5)
  Kaji (A) — FADING: 3 straight losses coming in, all vs ~1500 (Avg) fields:
    07-22 L Pareja(1500) 6-4 4-6 6-3 | 07-14 L Ma(1556) 6-3 6-3 |
    07-10 L Pareja(1500) 6-4 6-4 | 07-09 W McNeil(1500) grind | 07-07 W Johnson(1500)
    Several recent losses are blowouts (1-6, 0-6 sets) — blowout-prone vs class.
  Preston (B) — genuine tour class, mixed shape: 07-22 L Broadus(1500) 7-6 2-6 3-6 |
    06-23 Wimbledon QF L Grant(1627) | 06-18 Nottingham L Li | 06-15 W Blinkova(1676) |
    06-14 Nottingham FINAL L Ruzic(1682). Two clear classes above Kaji.

PATH / FATIGUE / NARRATIVE
  Both last played 07-22 (5 days rest); neither deep at Vancouver. No extreme
  fatigue spot — roughly equal. A age 31 / B age 20 (irrelevant, no penalty needed).
  Preston recently grass-heavy but played hard on 07-22 — minor transition, not
  material. No letdown (Kaji Taipei final 06-21 >7d; Preston Nottingham final 06-14
  >5wk). ELO agrees with ranking (A 338 weak, B 126 strong) — ELO credible.

MARKET-BY-MARKET (off corrected distribution, B 0.744)
  ML B at 1.30: implied ~0.778, my ~0.75 — no edge; min_odds 1.36 > 1.30 -> PASS.
  ML A at 3.90: no credible case; sim/ML biased. PASS.
  Total 19.5: over 0.51 / under 0.49 = coin flip; market over 1.92 / under 1.99
    prices the same. No edge either side. PASS.
  Hcap A +4.5 at 2.18: covers 0.553 -> ~8% raw edge, BUT abs(line)>=4.5 = high-var
    tier requiring >=10% edge; ~8% fails. Estimate also fragile (hinges on subjective
    correction point) and B -4.5 blowout is live given Kaji's 0-6/1-6 losses. PASS.

GATES (strongest candidate: A +4.5)
  G1 edge-origin: FAIL — ~8% raw vs 10% wide-AH bar; edge only exists at a
     subjectively-tuned sim point, not robustly from model-vs-market.
  G2 consensus:   FAIL — market+ELO both against A; no hard independent evidence
     Kaji stays within 4.5 (lone break-even ML + biased sim don't count).
  G3 fatal-risk:  FAIL — Kaji's recent losses include blowouts (1-6, 0-6) vs a
     genuinely higher-class opponent; B -4.5 blowout directly negates the +4.5 cover.

VERDICT: PASS. The market efficiently prices a real class gap that opponent-adjusted
ELO confirms; the sim/ML coin flip is an artifact of Kaji's weak-field-inflated
serve/return. No market offers a clean edge that survives the gates.