TennisEdge's AI model finds no betting value in T. Atmane vs A. Tabilo (Washington on 2026-07-29) — the market is efficient, with T. Atmane favoured at 1.60.
═══════════════════════════════════════════════
T. Atmane (A) vs A. Tabilo (B)
ATP Washington — hard — 1/8-finals
═══════════════════════════════════════════════
MODELS vs MARKET
ELO A 44.8% (B 55.2%)
ML A 49.9% (B 50.1%)
SIM A 53.5% (B 46.5%)
Book A 1.60 / B 2.35 → implied A 59.5%
BETFAIR (executable) A 1.90 / B 2.04 → vig-free A 51.8% / B 48.2%
The book quote is stale/soft — the exchange prices this as a near-pickem with
Atmane a whisker of a favourite. The brief's `value_pick B, edge 6.0%` is computed
off the 2.35 BOOK price; Betfair says 2.04. Per pricing discipline that KILLS the
edge, it does not confirm it — there is no 6% anywhere on the executable price.
Against the exchange the three models STRADDLE the market (ELO leans B, ML dead
level, sim leans A). Straddle = no directional disagreement to stake.
RECENT FORM (5 most recent each)
T. Atmane (24, rank 51)
07-28 W Tiafoe (1886, Elite) 6-4 3-6 4-6 GRIND ← over-performs bracket
07-01 L Tiafoe (1786) grass 4 sets
06-22 L Diallo (1684) grass 3 sets
06-17 L Medvedev (1836) grass straights
06-15 W Landaluce (1771) grass 3 sets + TB GRIND
→ 1-4 across the grass swing, clay before that near-winless. Then 27 days off
and an Elite scalp in match one at Washington. Shape: SHARPENING off a low base
(Tiafoe win is the only bracket-beating result in the window).
A. Tabilo (29, rank 35)
07-27 W Griekspoor (1814, Strong) 7-6 4-6 6-4 GRIND
07-22 L Torres (1629, Avg) straights ← UNDER-performing
07-18 L Rublev (1919, Elite) SF 3 sets, narrow neutral-to-positive
07-17 W Tirante (1500) (score n/a)
07-16 W Midon (1665) 3 sets
→ Bastad SF on clay, ugly straight-sets loss to a 1629, then a competent
clay→hard switch win over Griekspoor. Shape: STABLE.
PATH / SHARPNESS
Both players are exactly 1 match deep at Washington — no match-sharpness edge
either way. Both arrive off a surface transition (Atmane grass→hard w/ a 27-day
layoff; Tabilo clay→hard), and both have already absorbed it in round one.
Rust argument on Atmane is spent — he beat an 1886 in that first match.
FATIGUE (Step 2 calibration)
Atmane: 1 match/14d, 140 min, no back-to-back, last match 07-28 (3-set WIN).
Tabilo: 5 matches/14d, 734 min, back_to_back 2, last match 07-27 (3-set WIN).
BOTH sides won a 3-setter the day prior → the measured −4.5pp 3-set-hangover
haircut applies to BOTH and CANCELS. Applied once, not stacked. Ages 24 and 29
are both in the 23-29 baseline band — no age modulation on either side.
Tabilo's 594-minute cumulative court-time surplus is CONTEXT ONLY (large-sample
evidence: next-match win rate is ~unaffected by prior-match length). It is not
an edge and is not used to move a price. If anything the market has ALREADY
shaded Atmane from 1.60 (book) toward 1.90 (exchange) — the tired-Tabilo story
is priced, and then some.
NARRATIVE
Let-down: none (no Final for either player inside the window; Tabilo's Aix en
Provence title is 2026-05-03, ~86d ago — outside the 7d let-down horizon).
Venue momentum: neither player has prior Washington history in the window.
Surface transition: both — noted above, both already played through it.
ELO/ranking sanity: Tabilo rank 35 + ELO-favourite, Atmane rank 51 + ELO-dog.
ELO and ranking AGREE on both sides — no farmed-rating discount needed, ELO credible.
PRICING
B (Tabilo) @ Betfair 2.04 → eff 1.988 → need 55.3% to clear the 5% bar.
Blended read for Tabilo ≈ 50-51%. FAIL by ~5pp.
A (Atmane) @ Betfair 1.90 → eff 1.855 → need 58.9% to clear the 5% bar.
Blended read for Atmane ≈ 49-52%. FAIL by ~7pp.
alt_markets.handicap / totals are null and model margin probs are null — no
margin read available, so no cover or total to price. ML only, and ML is a PASS.
GATES
G1 edge-origin: FAIL — stripped of all fatigue/narrative, the raw model-vs-market
gap on the EXECUTABLE Betfair price is ~1-3pp on either side, far under the 5%
bar. The only 5%+ number in this match comes from a stale 2.35 book quote the
exchange contradicts at 2.04.
G2 consensus: FAIL — ML sits on the market (49.9% vs 51.8%) and the sim sits on
the other side of it from ELO. A lone ELO split with the sim opposing it is not
hard independent evidence.
G3 fatal-risk: FAIL — the one qualitative story here (Tabilo's 734-min load) is
precisely the visible-fatigue narrative the calibration says the market
over-prices, and Atmane's own 3-set hangover cancels it. Backing Atmane on it
would be paying for a story already in the 1.60→1.90 drift.
VERDICT: PASS
Near-pickem on the exchange, models straddling the price, both sides carrying the
same day-after 3-set toll, and the only apparent value is an artifact of a stale
book quote. Nothing to stake.
═══════════════════════════════════════════════