TennisEdge's AI model finds no betting value in B. Tomic vs K. Khachanov (Los Cabos on 2026-07-29) — the market is efficient, with K. Khachanov favoured at 1.25.
═══ B. Tomic (A) vs K. Khachanov (B) — ATP Los Cabos, hard, 1/8-finals ═══
MODELS vs MARKET
Book A 4.00 / B 1.25 → A 23.8% (5% overround, coherent)
Betfair ML A 4.10 / B 1.08 → sums to 1.17 = very wide/illiquid spread.
Exchange back-side on Khachanov (1.08 = 92.6%)
is SHORTER than the book (1.25 = 80%).
ELO A 24.8% → agrees with market
ML A 33.9% → lone dissent
Blend A 35.3% → the published value_edge (+11.5%) comes from here
Sim (base) A 59.9% → implausible; see SIM VALIDATION
Executable price for A = Betfair 4.10 → commission-adj 3.945 → 25.3% required.
min_odds at any defensible A-prob (≤25%) = 1.05/0.25 = 4.20 > 4.10 available.
SIM VALIDATION & BOUNDED RE-RUN (Step 4b)
Baseline inputs: Tomic spw .595 / rpw .419 (50 matches) vs Khachanov spw .590 /
rpw .405 (50 matches) → sim makes the world #194 a 59.9% favourite over the
world #17. Divergence vs ELO (24.8%) and market (24%) is enormous → inputs are
the problem, not the market.
Diagnosis from the match window: Tomic's 50 matches are Challenger/ITF —
opponent ELO 1355–1722 (Lincoln, Little Rock, Tyler, Bloomfield Hills, Los Cabos).
Khachanov's are Rome / Roland Garros / Wimbledon — opponent ELO 1593–1990
(Ruud 1990, Prizmic 1908, Trungelliti 1772, Van de Zandschulp 1761). The sim is
NOT opponent-quality adjusted, so Tomic's serve/return are fattened by weak
fields and Khachanov's are deflated by elite fields. Textbook input bias.
Bounded correction (max allowed ±0.05, logged):
spw_a 0.595 → 0.545, rpw_a 0.419 → 0.369 (re-run, neither clamped)
→ Tomic win prob 59.9% → 14.2%, expected margin +1.3 → −4.3 games.
Bracket: max-corrected 14.2% ... ELO 24.8% ... market 24%. True A-prob sits AT
or BELOW the market price. Blend of 35.3% is contaminated by the broken sim leg.
RECENT FORM (5 most recent each; classification used all fetched matches)
B. Tomic (33, #194) — hard, at this venue all week
07-28 W A. Hernandez (1424) 1-6 0-6 DOM [today, main draw R32]
07-27 L T. Boyer (1500) 0-0 ret/wo [row labelled "Final" — quali]
07-26 W M. Tobon (1390) 7-5 4-6 6.5-7.7 GRIND — 3 sets vs a 1390
07-21 L T. Boyer (1595) — Bloomfield Hills
07-19 W M. Lajal (1705) 4-6 7-6 7-6 GRIND — Lincoln title
Shape: FADING / UNDER-performing. Evidence: 3-set grind over a 1390 ELO
opponent on 07-26 and a two-tiebreak Challenger final — winning ugly against
weak fields. Level is Challenger, and he is grinding at that level.
TITLE: Lincoln 9d ago (hard). Let-down: lost a final 1d ago → NOTE only.
K. Khachanov (30, #17) — last match 07-04, 24 days idle
07-04 L F. Cobolli (1681) 6-0 6-7 7-6 2-6 2-6 5 sets, Wimbledon R32
07-02 W Y. Hanfmann (1610) 6-3 6-4 6-4 DOM
06-30 W B. Harris (1693) 6-3 5-7 6-3 6-3
06-16 L E. Quinn (1481) 6-1 4-6 4-6 grass wobble
05-27 W M. Trungelliti (1772) 4 sets, Roland Garros
Earlier clay swing is the real level read: beat Prizmic (1908) 6-1 7-6,
Van de Zandschulp (1761), Shevchenko (1680), took a set off Ruud (1990).
Shape: STABLE at a class Tomic has not faced in this window.
PATH / RUST / FATIGUE (Step 1b + Step 2)
Tomic: 4 matches at Los Cabos this week (back_to_back=4, matches_7d=4,
matches_14d=9, 1006 court minutes) and he played TODAY. Normally "match-sharp
beats rusty" — but this is an extreme spot: 3rd+ match on consecutive days,
plus a same-day/next-day turnaround. Age 33 (≥30) → harden one tier: real
disadvantage, not context. He also won a 3-setter 2d ago.
Khachanov: 0 matches in 14d, 24 days since Wimbledon, first match at the venue,
grass → hard TRANSITION. Genuine rust risk — depressed level and much wider
margin variance in match one. This is precisely why the book is 1.25 and not
1.15; the rust is ALREADY in the price, and the exchange still wants 1.08.
Note: the fatigue sits on the side the models want backed. Per Step 2, a
fatigue-drifted price is often the value side — but here the drift is on the
OTHER side (rust), so that inversion does not rescue Tomic.
DATA QUALITY
Tomic's Los Cabos rows carry inconsistent round labels ("Final" 07-27 with sets
"0-0", then "1/16-finals" 07-28) — reads as a qualifying final ret/wo followed
by main-draw entry. Noted, not leaned on.
ELO / RANKING SANITY
Tomic ELO ≈ rank (#194, Challenger results) → ELO credible and it says 24.8%.
Khachanov #17 with an ELO built on Rome/RG → credible. No divergence to exploit.
The only outlier in the whole stack is the unadjusted sim.
GATES
G1 edge-origin: FAIL — the +11.5% edge exists only because the blend carries a
sim leg that rates a #194 a 59.9% favourite over a top-20. Strip that leg and
ELO (24.8%) sits on top of the market (24.4%): no numeric gap to stake.
G2 consensus: FAIL — market AND ELO both against; only ML dissents. The
"hard independent evidence" on offer is live-event wins over 1424 and 1390
ELO opposition, which is not evidence against the world #17.
G3 fatal-risk: FAIL — class gap #17 vs #194 negates the thesis outright if
Khachanov simply holds serve, and the schedule burden (4th straight match day,
age 33) is on the side being backed. Units cannot buy that back.
MARKETS
alt_markets.handicap and .totals are null → nothing stakeable off the exchange
there. Corrected sim expects a −4.3 game margin for Tomic, but with no priced
line there is no bet to make.
Khachanov side: Betfair 1.08 → comm-adj 1.076 → 92.9% required. My read is
~80–85% (rust + transition + first match back). Negative edge. Book 1.25 is not
the executable price and cannot be used to manufacture one.
VERDICT — PASS
Three gates fail. The published edge is an artifact of opponent-unadjusted
serve/return inputs, and the exchange is pricing Khachanov shorter than the book
rather than softer. No side offers a stakeable price.