TENNISEDGE.GG

C. O'Connell vs A. Martin — Prediction & Analysis

Washington · · July 25, 2026 · 19:00 UTC

TennisEdge's AI model finds no betting value in C. O'Connell vs A. Martin (Washington on 2026-07-25) — the market is efficient, with C. O'Connell favoured at 1.19.

Surface hard Odds C. O'Connell 1.19 · A. Martin 4.30

Verdict: no bet

The model found no value at current prices — the market has this one priced about right. That verdict is part of the public record too.
Profiles: C. O'Connell · A. Martin · head-to-head record

Full analysis

C. O'CONNELL vs A. MARTIN — Washington (hard)
═══════════════════════════════════════════

MODELS & MARKET
  ELO:    75.6% A          ML model: 43.4% A          Blend: 66.1% A
  Book:   1.19 / 4.30 (implied 78.3% A)
  Betfair: 1.31 / 3.85 (implied 76.3% / 26.0%) — executable prices
  Value flag: B, +12.2% (driven entirely by the lone ML-model split)
  ELO vs ranking: A rank 150, B rank 269 — ELO-favourite status consistent with rank gap; no farmed-ELO flag.

DATA QUALITY WARNING
  matches_b shows "A. Martin" playing TWO tournaments on different continents on the
  same days (Bloomfield Hills, USA hard AND M25 Ollersbach, Austria clay, 07-21→07-25).
  Almost certainly two different players merged into one record. Martin's form history
  and the precomputed fatigue block (9 matches/14d, 1350 court min) are unreliable.

RECENT FORM (5 most recent shown)
  O'Connell: last match 06-25 (30 days off). Wimbledon run — L Final vs McDonald (1548-avg),
  W SF Tomic (Avg), W QF Schwaerzler (Weak, GRIND), plus Nottingham title 06-20 (W Final vs
  Virtanen, Strong, double-TB GRIND). Grass-heavy block → hard-court TRANSITION + 30d layoff rust.
  Martin (corrupted record): L M25 Ollersbach SF 07-25 (vs 1558, GRIND), L Bloomfield Hills QF
  07-24 (GRIND), W Bloomfield Hills R16 07-23, W M25 Ollersbach R16 07-23, W M25 Ollersbach R32
  07-21. All futures/challenger-level opposition; GRIND-heavy; losing at M25 level this week.

FATIGUE
  A: 0 matches/14d, well rested (age 32 — layoff cuts both ways: fresh but rusty; grass→hard transition).
  B: nominally 9 matches/14d, 1350 court min, played yesterday — but polluted by the merged record;
     cannot be trusted as a signal either way. Treated as unknown, not as an edge.

NARRATIVE
  A: TITLE Nottingham 35d ago (grass); Wimbledon Final loss 30d ago — too old for letdown. TRANSITION grass→hard.
  B: no venue history at Washington; Trieste Final loss 07-06. TRANSITION clay(?)→hard (last match clay).

GATES (candidate: Martin ML @3.85)
  G1 edge-origin: PASS — ML model 56.6% B vs exchange 26.0% is a numeric model-market gap, no narrative needed.
  G2 consensus:   FAIL — market AND ELO both sit against the pick; the lone ML-model split has zero hard
                  independent evidence (Martin lost yesterday at an M25 to a 1558-ELO player; record corrupted).
  G3 fatal-risk:  FAIL — the merged-player data means the model was likely trained/fed a polluted history;
                  if the ML split is an artifact of bad data, the thesis evaporates entirely.

  O'Connell side: Betfair 1.31 → effective 1.29, implied 77.3% vs blend 66.1% — negative edge, no bet.
  alt_markets handicap/totals: null — ML only.

VERDICT: PASS
  The only model-market disagreement comes from one model reading a provably corrupted opponent
  history. ELO + exchange agree O'Connell ~76%. No stakeable edge on either side.

More from Washington