A. Sabalenka vs M. Uchijima — Prediction & Analysis

TennisEdge's AI model finds no betting value in A. Sabalenka vs M. Uchijima (Toronto on 2026-08-04) — the market is efficient, with A. Sabalenka favoured at 1.02.

Model probabilities

Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.

We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?

Elo ratings

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 March 28, 2026.

Head to head

A. Sabalenka leads 1–0 over 1 meeting. Full record.

Recent form

All Aug. 4, 2026 results & picks →

Full analysis

═══ A. Sabalenka (A) vs M. Uchijima (B) — WTA Toronto, hard, 1/32-finals ═══

MODELS vs MARKET
  ELO       A 96.0%
  ML        A 84.2%  (implies B 15.8%)
  Model     A 96.0%
  Book      A 1.02 / B 15.0  (implied A 93.6%)
  Betfair   A 1.04 / B 15.0  → EXECUTABLE baseline
  Pinnacle  A 1.034 / B 15.04, no-vig A 93.57% (fetched 2026-08-03 22:55, ~1h old)
  Rank      A #1 vs B #107 — ELO and ranking agree on both sides; ELO is credible here.
  The only dissent is ML at 84.2%. Its documented failure mode is exactly this:
  regression toward break-even on extreme favourites. ELO, Betfair and Pinnacle all
  cluster 93.6–96.0%. Two sharp independent market reads within 0.03% of each other
  is about as tight a consensus as this slate ever produces.

RECENT FORM (max 5 printed per player)
  A. Sabalenka (28, #1) — last match 2026-07-05, 30 days ago
    07-05 Wimbledon 1/8  L vs N. Osaka (1811)      2-6 6-7   — GRIND loss, tight 2nd
    07-03 Wimbledon 1/16 W vs J. Ostapenko (1780)  6-4 6-4   — DOM over Strong
    07-01 Wimbledon 1/32 W vs M. Kessler (1688)    6-1 7-6   — GRIND (TB)
    06-29 Wimbledon 1/64 W vs T. Kostovic (1755)   6-2 6-3   — DOM over Strong
    06-20 Berlin SF      L vs J. Pegula (1865)     4-6 7-6 0-6 — decider collapse
    Also: 06-03 RG QF L vs Shnaider 6-3 5-7 0-6 — second 0-6 decider in the window.
  M. Uchijima (22, #107) — last match 2026-08-02 (yesterday)
    08-02 Toronto 1/64   W vs C. Bucsa (1744)      6-3 6-4   — DOM over Strong, MAIN DRAW
    08-01 Toronto Q-Final W vs H. Sakatsume (1667) 6-2 6-4   — qualifying final, DOM
    07-30 Vancouver QF   L vs M. Inglis (1657)     4-6 7-6 3-6
    07-29 Vancouver 1/8  W vs H. Kinoshita (1613)  2-6 6-4 7-5 — GRIND
    07-28 Vancouver 1/16 W vs M. Kuramochi (1452)  6-1 6-4
    DATA CAVEAT: B's history is contaminated — W15 Brisbane rows collide on the same
    dates as Vancouver rows, several "Won" rows carry losing scorelines, and a
    "Wimbledon QF" vs a 1500-ELO opponent is plainly mislabelled ITF. Treat B's
    aggregate stats as low-trust; only the two Toronto results are clean.

STEP 1b — SHAPE / PATH / SHARPNESS
  Sabalenka: FADING-ish. Two 0-6 deciders in the window (Pegula, Shnaider) and a
    straight-sets exit to Osaka. Genuine signal, but all of it is 30+ days stale.
  Sabalenka PATH: RUST RISK — 30 days since her last match, and grass → hard
    TRANSITION. First match at the venue, no bye-week reps. This is the real
    qualitative case for the dog.
  Uchijima: SHARPENING and MATCH-SHARP. Three wins at this venue this week
    (2 quali + a clean 6-3 6-4 main-draw win over Bucsa, ELO 1744). Conditions-
    adjusted, timing grooved. Best possible dog profile on paper.

STEP 2 — FATIGUE
  A: 0 matches/14d, fully rested. Rust, not fatigue.
  B: 11 matches/14d, 7 in 7d, 32 sets, 903 court minutes, back_to_back 5, played
    YESTERDAY. This is an extreme-spot load by any tier. Age 22 softens one tier
    (recovers fast), but yesterday's match was a 2-set win, so the calibrated
    3-set-hangover haircut (−4.5pp) does NOT apply — a 2-set win yesterday is
    roughly neutral. Net: B carries a mild real disadvantage; the market has NOT
    shaded for it (15.0 both venues), so this is context only, not an edge either way.

STEP 3 — NARRATIVE
  Let-down: B won a Final 08-01 — but it is the QUALIFYING final, not a main-draw
    title. No genuine let-down spot; do not read it as one.
  Venue momentum: B 3-0 at Toronto this week. A: no matches in window.
  Title recency: none for either in the window.
  Surface transition: A grass → hard = TRANSITION (level-depressing, widens margins).
    B is already on hard. Directionally this supports a B dog ML, not a B cover.

GATES
  G1 edge-origin: FAIL — Best dog price is Betfair 15.0, effective 14.30 (implied
    6.99%). Clearing the 5% bar requires my B probability ≥ 12.0%. Honest estimate
    for B is 8–9%: world #1, 30 days rust and a surface change knock Sabalenka's
    level down, but not to a 1-in-8 loss against a #107 qualifier who is 11 matches
    deep in 14 days. The only model that gets B above the bar is ML (15.8%), which
    is precisely the model with a known extreme-favourite regression bias and is
    contradicted by ELO, Betfair AND Pinnacle. Edge at my number = ~2%, well short.
  G2 consensus:   n/a (stake-tier only; no bet reaches the tier).
  G3 fatal-risk:  n/a (no margin bet — handicap/totals not quoted).

MARKET CHOICE
  alt_markets: handicap null, totals null, model null. Only betfair_ml is live.
  AH: not quoted
  ML A: Betfair 1.04 → effective 1.038 = 96.34% implied. My A number ~91–92%.
    min_odds = 1.05/0.92 = 1.14. 1.04 is far below the minimum → AUTOMATIC PASS.
    The favourite is priced through fair value; there is no world where laying
    96.3% on a rusty #1 is value.
  ML B: 15.0 → effective 14.30 = 6.99%. min_odds at p=0.09 is 11.67, so the price
    itself clears the minimum-odds screen — but the EDGE (2.0%) does not clear the
    5% bar. Price being long enough is necessary, not sufficient.
  No refutation check run — PASS decisions get no web access by design.

VERDICT — PASS
  This is the textbook shape of a match the market has right. The qualitative story
  is genuinely attractive on the dog side (rusty, fading, surface-switching #1 vs a
  match-sharp qualifier who just beat a 1744 in the main draw), and it is exactly
  the story I would want if the price were 25.0. At 15.0 it is already paid for.
  Two independent sharp reads (Betfair exchange and Pinnacle no-vig) sit inside
  0.03% of each other at ~93.6%, ELO is at 96%, and the lone dissenter is the model
  whose known bias points in exactly the direction of its dissent. Backing B needs
  me to believe a 12%+ upset rate; backing A needs me to accept 96.3% implied on a
  player who has not hit a competitive ball in a month. Neither is a bet. No stake.