TennisEdge's AI model finds no betting value in A. Tabilo vs T. Griekspoor (Washington on 2026-07-27) — the market is efficient, with T. Griekspoor favoured at 1.71.
A. Tabilo (29, #35) vs T. Griekspoor (30, #25) — Washington, hard, 1/16-finals
═══ MODELS vs MARKET ═══
ELO Tabilo 44.0% | ML 44.1% | Sim(blend) 51.0% | Book implied A 44.3%
Betfair ML: Tabilo 2.5 (40.0%) | Griekspoor 1.65 (60.6%). Book: 2.15 / 1.71.
value_pick A, edge 6.7% — but that edge is sim-vs-BOOK. ELO, ML AND the
exchange all agree Griekspoor is the ~58-60% favourite; the sim is the lone
outlier pulling A up. Ranking agrees with ELO (Griekspoor #25 > Tabilo #35).
═══ SIM RELIABILITY (resimulate) ═══
Raw sim: Tabilo win 72.2%, expected margin +2.8 A, A -2.5 covers 64.8%,
Under 23.5 60.1%. This diverges ~30pts from ELO/ML/market — inputs suspect:
- Tabilo spw .63/rpw .409 farmed ENTIRELY on clay (Aix title, Valencia SF,
Bastad SF); zero hard-court matches in window. Washington is hard →
clay-inflated serve/return applied to wrong surface.
- Griekspoor spw .606/rpw .388 polluted by 3 UTS EXHIBITION matches (point
scoring "9-19 14-13...", elo 1500 placeholders) + a poor grass losing run.
Both inputs contaminated → sim margin/win/totals distribution untrustworthy
here. No ±0.05 override rescues a 30-pt miss vs three converging signals.
═══ RECENT FORM (last 5) ═══
Tabilo (all CLAY/grass, NO hard): L Torres(1629) 4-6 4-6 (flat straight loss
to Avg, UNDER-perform) | L Rublev(1919,Elite) 6-4 4-6 6-4 (competitive) |
W Tirante(1500) no-data | W Midon(1665) 6-4 6-7 6-4 GRIND | L Majchrzak grass.
Shape FADING — most-recent is a limp straight-sets loss to a 1629 on clay.
Griekspoor (real tour form poor): UTS exhibitions x3 (discount) | L Duckworth
(1541,weak) grass 4sets | L Shimabukuro grass | L Zhang grass | W VDZandschulp.
Losing skid on grass, BUT career hard-courter, higher-ranked, ELO credible.
═══ PATH / SHAPE / SURFACE ═══
Both play their FIRST match at Washington (rust risk both sides). Tabilo carries
an added clay→hard TRANSITION (level depression + WIDER margin variance) — a
direct risk to any Tabilo cover or tight-margin thesis. Griekspoor's last real
tour match was 27d ago (Jun 30) with only UTS hard-court hits since.
═══ FATIGUE (Step 2) ═══
Tabilo 4m/14d, 576min, b2b2, last 07-22 (5d rest). Griekspoor 3m/14d but those
are UTS; last 07-18. Neither is an extreme spot; cumulative minutes = context
only, no edge. Age 29/30 baseline. Not a factor.
═══ NARRATIVE ═══
No first-Final let-down in window. No title-recency edge on the hard surface
(Tabilo's Aix title is clay, doesn't carry). Venue: neither has Washington
history to amplify anything.
GATES
G1 edge-origin: FAIL — staked edge exists only via the contaminated sim
(clay-farmed A, UTS-polluted B); vs credible ELO/ML 44% and the executable
Betfair 2.5 the edge is ~2.8%, below the 5% bar. No clean numerical edge.
G2 consensus: FAIL — market AND ML both against backing Tabilo; no hard
independent evidence (Tabilo has ZERO hard-court results in window; no
on-surface H2H). A lone surface-biased sim split does not pass G2.
G3 fatal-risk: FAIL — Tabilo's clay→hard transition negates both an A ML
thesis (level drop) and an A cover thesis (margin widening cuts both ways).
═══ VERDICT: PASS ═══
The only signal pointing to a bet is a simulation whose serve/return inputs are
demonstrably off-surface (Tabilo) and exhibition-polluted (Griekspoor). ELO, ML
and the Betfair exchange converge on Griekspoor; the executable A price (2.5) is
worse than the soft book quote (2.15) that manufactured value_edge — the
exchange contradiction kills it. Griekspoor side offers no edge either (1.65 →
break-even ~61.8% > any credible estimate). All three gates fail. No bet.