TennisEdge's AI model finds no betting value in K. Rakhimova vs H. Watson (Washington on 2026-07-25) — the market is efficient, with K. Rakhimova favoured at 1.57.
K. Rakhimova vs H. Watson — WTA Washington, Semi-finals, hard
MODELS & MARKET
ELO: 67.6% A | ML: 67.3% A | Sim: 59.4% A
Book: 1.57 / 2.25 (implied 58.9% A) | value_edge 0.5%
Betfair ML: 1.02 / 1.02 — invalid/unformed quote, unusable. No handicap/totals priced.
Book-priced regime applies: ML bet needs >=8% edge vs book.
DATA CAVEAT (dominant factor)
Neither player's match history includes ANY Washington matches — last visible match
Rakhimova 07-02 (Wimbledon), Watson 06-25. Both are in this SF, so each has ~3 unseen
hard-court wins this week. Fatigue block is zeroed and wrong (they have played all week).
The market has watched those matches; my form/fatigue read cannot.
RECENT FORM (visible, stale — 5 most recent each)
Rakhimova (24, #91): L Sakkari 1631 (grass, GRIND tiebreak) | W Kalinina 1620 (GRIND) |
L Selekhmeteva 1456 (grass, GRIND) | L Raducanu 1778 | W Dart 1635 (GRIND).
Visible window ~4-7; no DOM wins; mediocre form for a -150 favourite.
Watson (34, #354): L Timofeeva 1435 (final) | W Prozorova 1599 (GRIND) |
W Sherif 1343 (DOM vs Weak) | L Zhang 1637 (final) | W Haddad Maia 1687 (Strong).
Two finals reached in June; competitive veteran level.
FATIGUE
Backend fatigue precompute shows 0 matches 14d for both — contradicted by their presence
in this SF. Unusable; treated as unknown, no fatigue edge assigned either way.
Ages: Rakhimova 24, Watson 34 (would matter if data existed; it does not).
NARRATIVE
Let-down: none live (Rakhimova lost London final 48d ago; Watson lost a final 30d ago).
Venue: no Washington history in either feed. Title recency: none (no Won+Final).
Transition: both grass->hard on visible data, but superseded by unseen Washington matches.
ELO/rank sanity: Watson #354 with 34yo comeback profile and two June finals — ELO (32%)
likely understates her; market at 41% reflects her live 3-0 run here. Do not lean on the
raw ELO gap to back Rakhimova.
GATES
G1 edge-origin: FAIL — no executable exchange price (1.02/1.02 invalid); vs book 1.57 the
ELO/ML edge is 67.5% - 63.7% = 3.8% < 8% book-priced bar; sim edge ~0.5%.
G2 consensus: PASS — backing A would align with market and ML; no consensus violation.
G3 fatal-risk: FAIL — entire Washington run (both players, ~3 matches each) is missing
from the data; Watson's unseen live 3-0 at this event is exactly the evidence that could
negate the class thesis, and the market has it while I do not.
VERDICT: PASS. Two of three models like Rakhimova above the market, but the simulation sits
on the market, the only executable price is broken, the book edge (3.8%) is under the 8%
book-priced bar, and the form/fatigue inputs are blind to the week that matters.