TennisEdge's AI model finds no betting value in M. Frech vs A. Li (Washington on 2026-07-27) — the market is efficient, with A. Li favoured at 1.57.
═══ M. Frech (A) vs A. Li (B) — Washington, HARD, 1/16-finals ═══
MODELS vs MARKET
ELO A 0.496 | ML A 0.463 | blended model A 0.432
Book A 2.42 (0.413) | Betfair A 2.88 (0.347) | Betfair B 1.52 (0.658)
ALL THREE models favour Li (each gives Frech <0.50); disagreement with market
is magnitude only, not direction. Sharp exchange has DRIFTED Frech OUT (2.88 vs
book 2.42) — the exchange is MORE bearish on Frech than book AND models, the
classic tell that the model number (grass-inflated ELO) is the stale party.
RECENT FORM (last 5 each)
Frech (1-4, FADING):
06-29 Wimb grass L Kalinskaya 1676(Str) 6-7 4-6 (narrow)
06-22 BadHom grass L Osaka 1620(Avg) 6-4 6-1 (decisive loss to Avg)
06-16 Berlin grass L Lys 1544(Avg) 6-7 3-6 (loss to Avg)
06-15 Berlin grass L Zhang 1663(Str) FINAL 2-6 4-6 (lost Final decisively)
06-13 Berlin grass W Sasnovich 1639(Avg) 3-set TB GRIND (last win)
-> 0-3 since the Berlin Final; ~28 DAYS RUST (last match 06-29).
Li (1-4, also FADING):
07-16 Athens-Q hard L Korneeva 1500(Avg) 6-2 6-2 (decisive loss to Avg)
07-14 Athens-Q hard W Timofeeva 1500(Avg) 6-2 6-1 DOM
06-29 Wimb grass L Sonmez 1732(Str) 5-7 6-1 4-6
06-23 BadHom grass L Alexandrova 1764(Str) 1-6 7-6 1-6
06-19 Nott grass QF L Golubic 1698(Str) 3-6 6-2 3-6
-> lost badly in Athens qualifying; last match 07-16 (11d).
SHAPE / PATH
Frech: FADING (blowout losses to Avg-bracket Lys/Osaka), RUST (~4wk off),
TRANSITION grass->hard (level depressed, margins widen). No hard-court reps
in window. ELO ~0.50 vs rank 45 is grass-Final inflation — discount it.
Li: also FADING (thumped by 1500 Korneeva in Athens qual), but higher rank (29),
more hard-court/recent reps, and the market+model favourite. Rank 29 vs even-ish
ELO = ELO may slightly UNDERSTATE her — do not back Frech on an ELO gap that
isn't even there.
FATIGUE
Neither played in last 7d. Frech well-rested but rusty (28d). Li last match 11d
ago. No extreme spot for either -> fatigue is a non-factor here.
NARRATIVE
Frech: Berlin Final LOST 06-15 (>7d, no live let-down effect now). Grass->hard
TRANSITION = RISK on any Frech cover. No title recency.
Li: no let-down/title signal. No venue history for either at Washington.
EDGES (vs commission-adjusted Betfair)
Frech ML : model 0.432 vs impl 0.359 = +7.3% RAW -> negated qualitatively
Li ML : model 0.568 vs impl 0.669 = -10% -> no value
Frech +3.5: model 0.568 vs impl 0.548 = +2.1% -> below 5% bar; transition=cover risk
Li -3.5 : negative, high-variance lay tier (needs 10%)
Under 21.5: model 0.515 vs impl 0.508 = +0.7% -> below bar
Over 21.5 : negative
GATES
G1 edge-origin: PASS — raw Frech ML gap is +7.3% pre-narrative (numerically real).
G2 consensus: FAIL — market AND ML both favour Li; only a near-coin-flip ELO
splits for Frech; no on-surface H2H or live-event evidence, and
the sharp exchange fades Frech harder than the book.
G3 fatal-risk: FAIL — Frech's ~4wk rust + grass->hard transition + 0-3 fade is
exactly the "shows up flat, gets run over" risk that negates the
underpriced-Frech thesis. Units cannot buy this back.
VERDICT: PASS. The only nominal edge (Frech ML +7.3%) is a stale-ELO mirage that
the sharp exchange has already corrected against; fails G2 and G3. No alt market
clears its bar. A match between two fading players with no validated edge on either
side -> no bet.