TennisEdge's AI model finds no betting value in B. Nakashima vs T. M. Etcheverry (Washington on 2026-07-27) — the market is efficient, with B. Nakashima favoured at 1.52.
B. Nakashima (A, rk33, age24) vs T. M. Etcheverry (B, rk57, age27) — Washington, HARD, 1/16-finals
MODELS & MARKET
Brief (stale): ELO_a 0.546 | ML_a 0.483 | model_a 0.541 — near coin-flip, value_pick B (+8.3%).
Fresh resimulate (calibrated, 50 matches, current window) OVERRIDES the stale brief model:
A win 0.641 | B win 0.359. Inputs: spw_a .609/rpw_a .417 (hold 85%) vs spw_b .591/rpw_b .407 (hold 81.6%).
Distribution: exp margin 1.8, exp total 22.3.
Betfair (executable): A 1.47 (raw 68%), B 3.1 (raw 32%). Book 1.52/2.52.
=> Fresh sim (A 64%) and the exchange (A ~66-68% no-vig) AGREE. The apparent "B value" in the brief
came from the STALE sim (A 54%) and its stale AH line (B+3.5 covers 65.6%); the current sim erases it
(B+3.5 covers only 52.2%). ML is the lone low-A outlier and is noisy/break-even.
RECENT FORM (last 5 shown; all fetched used for counts)
Nakashima: 07-18 UTS Won Kyrgios (EXO/hard) | 07-18 UTS Won Miguel (EXO) | 07-17 UTS Won Cerundolo (EXO)
| 07-17 UTS Lost Miguel (EXO) | 07-02 Wimb GRASS Lost Struff(1682) 5-set thriller.
Real form: strong grass swing (Queen's SF, DOM 7-5 6-3 over De Minaur 1898), then a tight Wimb 5-set loss.
UTS is exhibition format — discounted, but gave him HARD-court reps 9d ago. Hard is his best surface; US home event.
Etcheverry: 07-24 Kitz CLAY Lost Bublik(1823) 6-3 6-4 SF | 07-23 CLAY Won Buse(1830) 6-2 7-5
| 07-22 CLAY Won Rodionov(1658) 6-2 7-6 | 07-15 Umag CLAY Lost Merida(1742) | 06-30 Wimb GRASS Lost Sonego.
Clay specialist, sharp CLAY run (Kitz SF) — but that form is surface-mismatched here.
SHAPE / PATH / TRANSITION
Nakashima: STABLE, hard is his surface, home. Grass→hard transition softened by UTS hard reps + hard being his best court.
Etcheverry: TRANSITION clay->hard (RISK). His good recent scorelines are all clay; hard record is thin/poor.
The surface switch depresses his level — this is precisely why the exchange & fresh sim discount him vs his clay-inflated ELO.
FATIGUE (context only — no extreme spot)
Etcheverry played 07-22/23/24 (Kitz), last match 3d ago — recovered, no yesterday/3-in-3 spot. Nakashima rested.
Fatigue is NOT a factor and does not enter any thesis.
MARKET EDGES (fresh-sim prob vs commission-adjusted Betfair)
B ML 3.1: 0.359 vs 0.334 -> +2.5% (below 5% bar; min 2.93)
A ML 1.47: 0.641 vs 0.691 -> -5.0% (no)
B +3.5 @1.76: covers 0.522 vs 0.581 -> -5.9% (no)
A -3.5 @2.2: covers 0.478 vs 0.467 -> +1.1% (high-variance tier needs 10%; no)
Under 22.5 @2.02: 0.518 vs 0.508 -> +1.0% (no) | Over 22.5 @1.9: -5.7% (no)
Best available edge = 2.5% (B ML), below the 5% bar. No market qualifies.
GATES
G1 edge-origin: FAIL — best numerical edge (B ML +2.5%) is below the 5% bar once the stale brief model is replaced by the calibrated resimulation.
G2 consensus: FAIL — fresh sim + ELO + exchange all favor A; only stale ML leans B. Lone ML split, no on-surface H2H or live evidence for B.
G3 fatal-risk: FAIL — Etcheverry clay->hard transition (level-depressing) directly negates any B under/cover/ML thesis; supports the market's A favoritism.
VERDICT: PASS. The brief's "8.3% B value" was a stale-model artifact. The current calibrated sim converges with
the exchange (Nakashima ~64-68%), the surface transition reinforces that read, and no market clears the 5% edge bar.
Textbook model+market consensus once the fresh sim is used — no edge either direction.