TennisEdge's AI model finds no betting value in J. Tjen vs A. Kalinskaya (Washington on 2026-07-30) — the market is efficient, with A. Kalinskaya favoured at 1.34.
═══════════════════════════════════════════════════════
J. Tjen (A, #40, 24) vs A. Kalinskaya (B, #24, 27)
WTA Washington — hard — 1/8-finals
═══════════════════════════════════════════════════════
MODELS vs MARKET
ELO Tjen 38.1%
ML Tjen 39.2%
Sim Tjen 42.0%
Book Tjen 3.20 → implied 31.3% (value_pick A, +12.5%)
Betfair ML A 1.12 / B 1.36 → 89.3% + 73.5% = 162.8% implied
→ SNAPSHOT INCOHERENT. B's 1.36 tracks the book (1.34)
and is credible; A's 1.12 is a broken/mis-mapped runner.
No trustworthy executable price on the value side.
All three models disagree with market on Tjen by ~7–11pp. Genuine
numeric gap → candidate. ELO/ranking coherent both sides (#40 at 38%
vs a #24 — no farmed-rating flag).
RECENT FORM (5 most recent each; classification uses full window)
Tjen — 4W-8L in window
07-27 Washington HARD W vs Sramkova 1681 (Strong) 6-4 6-2 DOM
07-01 Wimbledon grass L vs Kasatkina 1652 TB+1-6 4-6
06-29 Wimbledon grass W vs Fernandez 1645 GRIND (TB)
06-23 Eastbourne grass L vs McNally 1732 3 sets
06-15 Nottingham grass L vs Maria 1757 straight
→ Grass swing was a wreck (5 losses in 6). One hard match, and it
was a clean DOM over a Strong opponent.
Kalinskaya — 9W-6L in window
07-28 Washington HARD W vs Kasatkina 1755 (Strong) 6-4 6-1 DOM
07-03 Wimbledon grass L vs Bencic 1876 (Elite) 6-4 4-6 7-6 (narrow)
07-01 Wimbledon grass W vs Parry 1730 GRIND (TB)
06-29 Wimbledon grass W vs Frech 1476 GRIND (TB) — under-perf
06-24 Bad Homburg L vs Ruse 1803 5-7 2-6
→ Plus a French Open QF run: DOM over Boisson 1759, beat Potapova
1949 and Osorio 1729; Rome win over Bencic 1801.
SHAPE / PATH / SHARPNESS (Step 1b)
Tjen — shape STABLE-to-improving on the surface switch: the 6-4 6-2
over Sramkova resets a bad grass block, but n=1 on hard.
Kalinskaya — SHARPENING: 6-4 6-1 over Kasatkina (1755) is DOM over
Strong, better than Tjen's DOM over 1681, and it happened at THIS
venue on THIS surface one day ago.
Path — both 1 match deep at Washington, both already absorbed the
grass→hard transition in R1 (most-recent surface = hard for both,
so no live TRANSITION flag for either). No rust for either.
Live this-event evidence points at Kalinskaya, not Tjen.
FATIGUE (Step 2)
Tjen 1 match/14d, 68 min, last 07-27 (2d), b2b 0
Kalinskaya 1 match/14d, 64 min, last 07-28 (1d), b2b 0
No extreme spot on either side. Kalinskaya played yesterday but in
2 sets / 64 min — the calibrated −4.5pp hangover applies only to a
3-SET win yesterday, so NO haircut here (a 2-set win is ~neutral and
already inside what the models price). Both 24/27 → baseline age.
Verdict: roughly equal. Fatigue is not an input to this match, and
the price has not drifted on any fatigue narrative.
NARRATIVE (Step 3)
Let-down: none — no Final in either window.
Venue momentum: neither has Washington history beyond this week.
Title recency: none.
Surface transition: none live (both played hard in R1).
ALT MARKETS
handicap null, totals null, model null → nothing priced to stake.
No margin market available, so no cover/total thesis is possible.
GATES
G1 edge-origin: [PASS] — numeric pre-narrative: ELO 38.1/ML 39.2/sim 42.0 on Tjen vs market 31.3% book (~+9.7% comm-adj, min 2.50); no fatigue/narrative in the edge.
G2 consensus: [PASS] — stake-tier only: 3 models with Tjen, BUT live this-event evidence runs against it (Kalinskaya DOM'd a 1755 yesterday) → contaminated signal, capped 0.5–1.0u tier.
G3 fatal-risk: [PASS] — ML bet, G3 never blocks; the negating risk would ride as risk_flag: at 1.0u.
VERDICT — PASS
Two independent reasons, either sufficient:
1) PRICE. Pricing Discipline #1: the edge must be taken against the
executable exchange price. The exchange quotes Tjen at 1.12 —
nonsense against B's credible 1.36 (163% book), so there is no
verifiable exchange price on the side with the edge. A book quote
the exchange cannot corroborate is a stale quote, not value. Even
under #4 (treat betfair_ml as absent → 8% bar vs book, which the
9.7% clears) the rule says prefer PASS in exactly this segment:
thin/unreliable exchange liquidity is where the backtest bleeds.
2) QUALITATIVE. A model–market gap needs independent qualitative
validation, and here the qualitative read validates the MARKET.
Kalinskaya is the stronger player on live evidence: #24 vs #40,
an FO quarter-final with wins over 1949/1759, and a cleaner DOM
(over 1755) than Tjen's (over 1681) at this event this week.
Tjen's model edge rests on hard-court rating while her live record
is 4-8 with a five-loss grass block and a one-match hard sample.
A skeptic passes this: nothing here is information the exchange
is slow to price.
Forecast winner Kalinskaya ~63%; my Tjen estimate ~37%, i.e. BELOW
the sim's 42% after the live-evidence read — which removes the edge
even before the price problem. No stake.
═══════════════════════════════════════════════════════