K. Rakhimova vs K. Siniakova — Prediction & Analysis

TennisEdge's AI model finds no betting value in K. Rakhimova vs K. Siniakova (Toronto on 2026-08-04) — the market is efficient, with K. Siniakova favoured at 1.50.

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 Aug. 2, 2026.

Head to head

K. Siniakova leads 2–1 over 3 meetings. Full record.

Recent form

All Aug. 4, 2026 results & picks →

Full analysis

K. Rakhimova (A, 24, #91) vs K. Siniakova (B, 30, #39) - WTA Toronto, hard, 1/32-finals

═══ MODELS vs MARKET ═══
ELO prob A      33.77%
ML prob A       41.77%
Composite       33.77%
Betfair ML      A 2.80 / B 1.52  -> implied A 35.7%
Pinnacle no-vig A 34.95% (fetched 22:55Z, ~3h old - fresh)
Book quote      A 2.60 / B 1.50 (context only; exchange is the baseline)

ELO, Pinnacle's sharp no-vig line and Betfair all cluster at 34-36% for
Rakhimova. Only the ML model dissents at 41.8%. That is a lone-model split
against the three sharpest estimates available, and it is worth 4.87% edge -
under the 5% bar - even if taken at face value.

ELO vs ranking: Siniakova #39 / ELO-favourite, Rakhimova #91 / ELO-dog. Rank
and ELO agree on both sides. No farmed-rating or understated-rating flag; the
ELO is credible here, which makes the ML model the odd one out rather than the
insight.

═══ RECENT FORM (5 most recent each) ═══
RAKHIMOVA (A)
  08-02 Toronto     hard  R64  W  V. Williams (1320, Weak)     6-4 6-1   DOM
  07-25 Washington  hard  SF   L  H. Watson (1594, Avg)        6-7 6-7   GRIND
  07-02 Wimbledon   grass R32  L  M. Sakkari (1631, Strong)    3-6 6-0 6.7-7.10
  06-30 Wimbledon   grass R64  W  A. Kalinina (1620, Strong)   6-4 4-6 5-7  GRIND
  06-20 Eastbourne  grass SF   L  O. Selekhmeteva (1456, Weak) 6-7 6-2 1-6

SINIAKOVA (B)
  07-01 Wimbledon   grass R32  L  N. Bartunkova (1782, Strong) 2-6 4-6
  06-29 Wimbledon   grass R64  W  Q. Zheng (1666, Strong)      6-4 6-4   DOM
  06-21 Bad Homburg grass R16  L  L. Samsonova (1754, Strong)  6-3 7-5
  06-17 Berlin      grass R16  L  J. Pegula (1811, Strong)     6-2 6-4
  06-16 Berlin      grass R32  W  R. Masarova (1803, Strong)   6-2 6-4   DOM

═══ SHAPE, PATH & SHARPNESS ═══
Rakhimova: STABLE-to-mildly-positive. The 6-4 6-1 DOM is over a 1320 opponent -
weak-field evidence, not a level statement. The more informative line is the
Washington semi-final, a double-tiebreak loss to a 1594: competitive, but
grinding out tight sets against Average opposition is not over-performance. Net:
solid hard-court sharpness, no upward break in level.
Path: 2 matches at this venue-week cycle, won at Toronto two days ago.
MATCH-SHARP - conditions-adjusted and timing grooved.

Siniakova: STALE, unreadable. Her entire window is grass, ending 07-01. Within
it, shape was neutral (DOM over Zheng 1666 and Masarova 1803; straight-sets
losses to Bartunkova/Samsonova/Pegula, all Strong). No hard-court evidence at
all in the window.
Path: last recorded singles match 07-01 - 33 DAYS AGO, and her first match at
this venue. Clear RUST RISK: depressed level plus widened margin variance in
match one.

This rust-vs-match-sharp asymmetry is the one genuinely live qualitative angle
in the match, and it points at Rakhimova. It is also exactly the sort of thing
G1 forbids from manufacturing an edge: it is narrative, not a numeric
model-market gap. Noted, not banked. (Caveat: Siniakova is an elite doubles
player whose doubles court-time is invisible to this dataset, so the layoff may
overstate her actual rust.)

═══ FATIGUE ═══
Rakhimova: 2 matches / 14d, 162 min total, last match 08-02 (2 days ago,
straight sets, ~81 min). No back-to-back. Age 24 = baseline recovery. NOT a
3-set winner yesterday, so the calibrated -4.5pp hangover haircut does NOT
apply. No extreme spot.
Siniakova: 0 matches / 14d. Fully rested - but per Step 1b that reads as rust,
not advantage.
Verdict: fatigue is a non-factor both ways. Cumulative load is context only and
is not permitted to price this match.

═══ NARRATIVE ═══
Let-down: None (no Final in either window at short range).
Venue momentum: Rakhimova 1-0 at Toronto this week; Siniakova no record.
Title recency: none in window for either.
Surface transition: BOTH players last played on grass before this hard swing -
but Rakhimova has already re-acclimatised (Washington + Toronto on hard),
Siniakova has not played a hard court in the window at all. Transition risk sits
entirely on Siniakova, which widens her margin variance in both directions - a
reason to avoid any tight margin bet here, not a reason to back one.

═══ MARKET CHOICE ═══
Sim validation: the baseline simulation returns Rakhimova 50.3% to win - a
14-17pp divergence from ELO (33.8%), market (35.7%) and Pinnacle (35.0%)
simultaneously. Divergence from every other estimate = suspect inputs, not
discovered value.
Inputs inspected: baseline A spw .577 / rpw .430 vs B spw .583 / rpw .424 -
near-identical, yet hold% is .618 (A) vs .703 (B), an internal inconsistency
that flags mismatched opponent pools. Rakhimova's 50-match sample is padded with
weak fields (V. Williams 1320, Gorgodze 1436, Selekhmeteva 1456, small grass
events); Siniakova's is almost entirely main-tour opposition at 1750-1850
(Boisson 1827, Bouzkova 1824, Masarova 1803, Pegula 1811, Waltert 1756). A's
serve/return numbers are inflated, B's deflated - the sim is opponent-unadjusted
and therefore biased toward A.
Bounded correction (logged): spw_a .577 -> .565, rpw_a .430 -> .415 (both well
inside the +/-0.05 clamp), shading A's stats for the weak-field inflation. The
re-run returns A 36.1% - converging with ELO, Pinnacle and Betfair. That is the
anchor distribution; all markets below are priced off this ONE run, so ML, cover
and total are internally consistent.
Note the sim's fragility: 3pp of SPW swings win prob 31pp (a -0.032 shade took A
to 18.8%). A margin tool that unstable does not deserve a staked cover.

Priced against commission-adjusted (0.95) exchange odds:
  ML A  @ 2.80  p=.361  edge -0.80%   (min 2.91)  FAIL
  ML A  @ 2.80  p=.4177 edge +4.87% on the lone ML model alone - still under bar
  ML B  @ 1.52  p=.639  edge -3.03%   (min 1.64)  FAIL
  AH A +3.5 @ 1.97  p=.5412 edge +2.08%  (min 1.94)  FAIL
  AH B -3.5 @ 1.93  p=.4588 edge -7.21%  FAIL
  TOT Over  21.0 @ 1.93  p=.500 edge -3.09%  FAIL
  TOT Under 21.0 @ 2.02  p=.500 edge -0.79%  FAIL
AH: A +3.5 @ 1.97
Best available edge is +2.08% (A +3.5) off the corrected distribution, or
+4.87% on the uncorroborated ML model. Nothing reaches 5%. No refutation check
run - PASS decisions get no web access.

═══ GATES ═══
GATES
  G1 edge-origin: FAIL — best pre-narrative exchange edge is +2.08% (A +3.5) / +4.87% on the lone ML model; both under the 5% bar. Only the rust-vs-sharp narrative would lift this, and narrative cannot create edge.
  G2 consensus:   stake-tier only (never blocks) — lone-model split (ML alone vs ELO+Pinnacle+Betfair) AND contaminated inputs (A's serve/return farmed on 1320-1456 fields), so this would cap at 0.5-1.0u `lone_signal:` even had G1 cleared.
  G3 fatal-risk:  FAIL for margin bets (hard veto) — the A +3.5 cover thesis has no margin-specific evidence: no on-surface H2H scorelines, A's own recent hard-court margin distribution is two matches deep, and Siniakova's 33-day layoff widens margins in BOTH directions, which cuts against a tight cover as readily as for it. N/A for ML (never blocks).

═══ VERDICT: PASS ═══
Three of the four sharpest estimates available - ELO, Pinnacle's no-vig line and
the Betfair exchange - agree inside 2pp on Rakhimova at ~35%. The one dissenter
is the ML model, and its dissent is worth less than the minimum bar even taken
uncritically. The simulation that appeared to offer a big edge was demonstrably
biased by weak-field serve/return inputs, and correcting it collapsed the edge
back onto the market price. The genuine insight here - Siniakova 33 days idle
and playing her first match at the venue against a match-sharp opponent who won
here two days ago - is real, but it is precisely the narrative G1 exists to stop
me banking. The honest read is that the market has already priced a 2.80 dog
about right. No bet.