TennisEdge's AI model finds no betting value in A. Mannarino vs L. Tien (Washington on 2026-07-27) — the market is efficient, with L. Tien favoured at 1.23.
A. Mannarino (38, #45) vs L. Tien (20, #26) — Washington, hard, R32.
MODELS (Tien win prob)
ELO: 79.2% | Sim(point-by-point): 82.8% | Betfair implied: ~77.5% | ML: 58.3% (outlier)
Three independent methods (surface ELO, serve/return sim, exchange) converge ~78-83%
on Tien. ML at 58% is the lone dissenter — almost certainly over-weighting ranking
proximity (45 vs 26) while ignoring that Tien is a legit top-30 talent and Mannarino a
fading 38yo. ML is the wrong model here; the consensus is right.
PATH / SHAPE
Tien: Geneva CLAY title late-May (beat Tsitsipas, Bublik 1869, Navone 1886), RG R16.
Genuine rising force. Grass season mediocre (Halle R16, Wimbledon R2). Last match
01-Jul → 26 days idle: RUST RISK. Transition grass→hard.
Mannarino: Hertogenbosch grass SF mid-June, Newport R16 loss (grass) last out 09-Jul,
18 days idle. 38yo, declining. Transition grass→hard. Historically fine on hard but
current level does not threaten the consensus.
Both first match at venue, both rusty, both transitioning off grass → roughly neutral.
FATIGUE: both matches_14d=0, well_rested. No extreme spot. Non-factor.
NARRATIVE: Tien's Geneva title (05-23, clay) is 2 months stale — live hard-court read,
not a letdown/venue signal. Nothing actionable.
MARKET SCAN (all executable Betfair prices, commission-adjusted)
ML Tien @1.29: edge +4.4% (sim) / +0.8% (ELO) — BELOW 5% bar; ML dissents heavily → no.
ML Mannarino @4.2: only ML (outlier) supports vs ELO 20.8% + sim 17.2% + market ~23%.
Lone-model split against full consensus. No.
Handicap 4.5: Mann +4.5 covers 46.2% @1.64 → edge -16%. Tien -4.5 covers 53.8% @2.0
→ edge +2.6%, wide/high-variance line needs +10% → no.
Totals 22.0: sim under ~63% @1.78 → nominal edge +5.5% (only market clearing min-odds).
Over @1.93 → -16%.
SIM DETAIL (baseline resimulate, 50 matches each, surface=hard)
Mannarino spw .574 rpw .404 hold .802 | Tien spw .618 rpw .431 hold .768
Win prob 17.2/82.8. Expected margin -3.9 (Tien by ~4). Expected total 20.3 CALIBRATED
(raw 23.3, a large -3.0 game bias correction). Inputs built on grass/clay only — zero
recent hard-court data for either player, so margin/total inputs are off-surface.
No justified single-input correction (both off-surface equally) → trust baseline, no re-run.
GATES (candidate = Totals Under 22.0, the only market clearing min-odds)
G1 edge-origin: PASS — 5.5% edge is numerical (sim calibrated dist vs market line), no narrative.
G2 consensus: PASS — market itself leans under (1.78<1.93); no consensus against.
G3 fatal-risk: FAIL — BOTH players are extreme tiebreak magnets. Mannarino recent:
6.5-7.7, 6.3-7.7, 6.5-7.7, 6.4-7.7, 6.7-7.9. Tien recent: 7.7-6.4, 6.5-7.7, 6.4-7.7,
6.2-7.7, 7.7-6.5. Both hold well and drag sets to TBs → matches run LONG. This directly
negates the under, whose thin edge exists ONLY via an unverifiable -3.0 bias correction
on off-surface inputs. Units cannot buy back a negating risk.
VERDICT: PASS. Efficient match — ELO, sim and the exchange all agree Tien ~80%; ML is a
lone artifact. Tien ML too short (edge <5%, fails bar). Every alt market is negative,
below-bar, or (the under) killed by G3. No priced edge to stake.