S. Bejlek vs I. Swiatek — Prediction & Analysis

TennisEdge's AI model finds no betting value in S. Bejlek vs I. Swiatek (Toronto on 2026-08-04) — the market is efficient, with I. Swiatek favoured at 1.11.

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 July 24, 2026.

Head to head

I. Swiatek leads 1–0 over 1 meeting. Full record.

Recent form

All Aug. 4, 2026 results & picks →

Full analysis

S. BEJLEK vs I. SWIATEK — WTA Toronto, hard, 1/32-finals
═══════════════════════════════════════════════════════

MODELS vs MARKET (player A = Bejlek)
  ELO        20.64%   <- lone dissenter
  ML         12.48%
  Sim (raw)  38.45%   <- implausible, see input audit
  Pinnacle no-vig 13.73% (fetched 2026-08-03 22:55, ~1h fresh)
  Betfair ML A 7.4 / B 1.13 -> A eff. 7.08 (5% comm) = 14.12% implied
  Book 6.75/1.11. Value_pick A +6.5% is ELO-driven only.

SIM INPUT AUDIT (Step 4b) — baseline sim is contaminated
  Baseline serve/return: Bejlek spw .576 / rpw .419 vs Swiatek spw .586 / rpw .432.
  The tool hands the world #2 essentially the SAME serve/return profile as a #37 —
  because inputs are unweighted, opponent-unadjusted averages. Bejlek's 28-match
  window is fattened by Athens QUALIFYING (Morvayova 1500, Tagger 1500) and clay;
  Swiatek's 50-match window is depressed by an elite-only field (Pegula 1933,
  Osaka 1834, Kostyuk 1953, Svitolina 1960, Eala 1857). Result: sim win prob 38.5%
  for Bejlek — 24pp above the market and 26pp above ML. Divergence, not convergence.
  CORRECTION + RE-RUN (logged): rpw_a 0.419 -> 0.370 (-0.049, within clamp; the only
  input I can justify numerically — Bejlek's return numbers built vs sub-1500 quali
  servers). Re-run: A win 17.56%, expected margin -3.5, A covers +6.5 = 81.0%.
  Sensitivity check, also shading spw_b 0.586 -> 0.630 (Swiatek serve vs elite-only
  field): A win 6.9%, A covers +6.5 = 70.7%. Note the incoherence — a 6.9% underdog
  still "covers" +6.5 seven times in ten. The margin engine is compressed and is NOT
  a usable anchor here in EITHER direction. Win-prob anchor taken from the single
  correction (17.6%); margin distribution rejected.

RECENT FORM (last 5 each)
  Bejlek (20, rank 37) — ELO 1743
    08-02 W  Xiy. Wang (1689, Strong)  3-6 6-4 5-7   GRIND, 3 sets
    07-24 L  L. Tagger (1708)          straight sets
    07-23 W  M. Timofeeva (1786, Strong) 1-6 1-6     DOM
    07-21 W  A. Blinkova (1601, Avg)   6-7 6-3 1-6   GRIND
    07-17 L  C. Tauson (1792)          6-2 6-3       straight-set loss
  Swiatek (25, rank 2) — ELO ~2035
    07-04 L  A. Eala (1857, Elite)     7-6 6-2
    07-02 W  Ka. Pliskova (1735)       1-6 3-6       DOM
    06-30 W  T. Townsend (1537, Weak)  1-6 6-2 3-6   GRIND (3 sets vs Weak)
    06-24 L  E. Navarro (1804, Elite)  5-7 6-2 3-6
    05-31 L  M. Kostyuk (1953, Elite)  7-5 6-1

STEP 1b — SHAPE & PATH
  Bejlek: STABLE-to-mixed. DOM over Timofeeva (1786) is the one over-performance;
    against that, a labored 3-set win over Wang and a straight-set loss to Tauson.
    PATH: 1 match at this venue, already through the hard-court transition (Athens
    hard mid-July) — match-sharp, and the Toronto win is live on-surface evidence.
  Swiatek: FADING-to-neutral on the last window (needed 3 sets vs Townsend 1537;
    R16 exits at both RG and Wimbledon), but her opponents are Elite-heavy so the
    losses read neutral. PATH: RUST RISK — last match 07-04, 31 days idle, first
    match at the venue. This is the standard post-Wimbledon reset the market prices
    every year for every top player, not private information.
  H2H: Swiatek won 6-2 6-3 (clay, 05-27, 10 weeks ago) — margin 7 games.

FATIGUE (Step 2)
  Bejlek: 4 matches/14d, 398 court min, last match 08-02 (2 days ago) — the 3-set
    win was NOT yesterday, so the calibrated -4.5pp hangover haircut does NOT apply.
    Age 20 (<=22, fast recovery) softens the cumulative load a further tier -> context
    only, no numeric effect. Swiatek: 0 matches/14d, fully rested.
  No extreme spot on either side. Per calibration, cumulative minutes alone is noise.

NARRATIVE (Step 3)
  Let-down: none (no Final in either window). Venue momentum: Bejlek 1-0 at Toronto
  this week. Title recency: none. Surface transition: Bejlek already converted to
  hard (Athens 07-13/17, Toronto 08-02); Swiatek grass -> 31d gap -> hard, which
  depresses level AND widens margin variance in match one.

GATES
  G1 edge-origin: FAIL — corrected-sim anchor 17.6% and a sane blend of all four
    estimates (~16%) give only +3.4% / +1.9% vs the executable 14.12% (Betfair 7.4,
    comm-adj 7.08). Only raw ELO 20.64% clears 5%, and ELO is contradicted by ML
    (12.5%), Pinnacle no-vig (13.7%) and the exchange. Min odds at p=0.176 is 5.98,
    but the absolute edge bar is not met pre-narrative. No stakeable edge exists.
  G2 consensus:   PASS (tier only) — lone-model support (ELO), so any bet would have
    been capped 0.5-1.0u with a lone_signal: prefix. Moot given G1.
  G3 fatal-risk:  FAIL (margin markets only) — the only margin bet the data would
    suggest is Swiatek -6.5, and it self-negates: her 31-day layoff + first match at
    the venue widens margin variance exactly when the thesis needs a clean blowout
    (her own last first-match-back went 3 sets vs Townsend, 1537). Hard veto per the
    handicap scope.

MARKET CHOICE
  ML A @ 7.4 (eff 7.08): my 16-17.6% vs 14.12% implied = +1.9 to +3.4% -> below the
    5% bar. PASS. ML B @ 1.13 is obviously no value.
  Handicap: A +6.5 @ 1.86 (53.8% implied). Sim says 81%/71% covers — but Swiatek's
    OBSERVED scoreline distribution says the opposite: only 1 of her 9 wins in the
    window finished within 6 games (margins 8, 12, 4, 7, 9, 9, 9, 11, 9), and she
    beat Bejlek by exactly 7 in the H2H. Bottom-up estimate for A covering +6.5 is
    ~40%, i.e. the sim and the evidence point in OPPOSITE directions across the
    market's own line. When the margin tool is incoherent with observed margins there
    is no trustworthy cover estimate — no bet either way; and the lay side is
    G3-vetoed above. PASS.
  Totals: line 18.0, over 1.88 (53.2%) / under 2.04 (49.0%). Corrected sim gives
    over 18.5 = 51.3%, the sensitivity run far lower — the estimate swings ~13pp on
    an input I cannot pin down. No stable edge. PASS.
  AH: A +6.5 @ 1.86 (quoted; not staked)
  alt_markets present (fetched 2026-08-03 23:07) — exchange ML and lines usable.

VERDICT: PASS
  The whole "edge" is one model (ELO) disagreeing with ML, Pinnacle and the exchange,
  plus a rust story the market prices for every top player every August. The sim that
  would have made this a big bet is demonstrably contaminated — it gives a #37 the
  same serve/return profile as the world #2 — and once its return input is corrected
  it lands at 17.6%, right on top of the market. Bejlek is match-sharp and Swiatek is
  idle, but 10 weeks ago Swiatek beat her 6-2 6-3 and Bejlek's hard-court evidence is
  qualifying-level wins plus a straight-set loss to Tauson. Nothing here is worth
  14-cent money at 7.4. No bet.