J. Lehecka vs V. Kopriva — Prediction & Analysis

TennisEdge's AI bot is backing V. Kopriva (moneyline) at 5.00 in J. Lehecka vs V. Kopriva (Montreal on 2026-08-06).

Model probabilities

Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.

We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?

Elo ratings

Our own ratings, computed from match results since 2018. Players with fewer than 5 rated matches are omitted rather than shown at a provisional starting value. Last updated July 30, 2026.

Head to head

J. Lehecka leads 1–0 over 1 meeting. Full record.

Recent form

All Aug. 6, 2026 results & picks →

Full analysis

J. LEHECKA vs V. KOPRIVA — ATP Montreal, 1/32-finals, HARD
RE-ANALYSIS (prior probe: Kopriva ML @4.60, 0.5u — exchange has drifted OUT to 5.00)

═══ MODELS vs MARKET ═══
ELO       A 71.9% / B 28.1%
ML        A 76.1% / B 23.9%
Blend     A 73.1% / B 26.9%
Sim       A 65.3% / B 34.7%  (baseline serve/return)
Book      1.17 / 4.10   |   Betfair ML 1.20 / 5.00 (exec venue)
Betfair no-vig: A 80.6% / B 19.4%. Commission-adjusted B price = 1+(4.00x0.95) = 4.80 -> breakeven 20.83%.
Every model prices Kopriva ABOVE the market's 19.4%: sim +15.3pp, ELO +8.7pp, blend +7.5pp, ML +4.5pp.
Pinnacle: null (Montreal R2 not posted yet — absence of data, not evidence).

═══ SIM VALIDATION & RE-RUN (Step 4b) ═══
Baseline inputs: Lehecka spw .613 / rpw .408 / hold .849 (50 matches); Kopriva spw .582 / rpw .411 / hold .748 (49 matches).
Divergence check: sim (A 65.3%) sits BELOW ELO (71.9%) and far below market (80.6%) — inputs suspect. Whose? Kopriva's rpw .411 is essentially identical to Lehecka's .408, which is implausible: 8 of Kopriva's last 15 matches are clay and only ONE is hard, so his return number is built on slow courts against mixed fields, not against a big first serve on Montreal hard.
Re-run logged: rpw_b 0.411 -> 0.385 (-0.026, within +/-0.05 clamp). Result: B 34.7% -> 23.3%, expected margin 2.0 -> 3.4 games.
Read: an 11pp swing in win prob from 2.6pp of RPW = the sim is on a knife-edge here and is NOT a robust anchor. I bracket rather than adopt — honest Kopriva range 23.3%–34.7%, and I take the LOW-MIDDLE (see ranking cross-check).

═══ RECENT FORM (5 most recent each) ═══
LEHECKA (24, rank 19)
  07-30 Los Cabos (hard) R16  LOST  C. Wong 1701      6-1 3-6 4-6   — led a set 6-1, collapsed. UNDER-performance.
  07-07 Wimbledon (grass) R16 LOST  Zverev 1891       4-6 5-7 6-3 6-7 — competitive vs Elite, neutral-positive
  07-04 Wimbledon R32         WON   Munar 1687        6-4 6-4 4-6 6-4 — solid win over Strong
  07-02 Wimbledon R64         WON   Molcan 1594       6-3 6-2 6-4   — DOM over Avg
  06-30 Wimbledon R128        WON   Popyrin 1534      6-4 6-2 6-4   — DOM over Avg
  SHAPE: STABLE-to-FADING. Grass block was genuinely good; the ONLY hard-court data point in the window is the Wong loss (6-1 3-6 4-6 to a 1701) — that is the surface that matters here, and it is under-performance.
KOPRIVA (29, rank 55)
  08-03 Montreal (hard) R64   WON   Galarneau 1626    6-4 6-4       — DOM over Avg, at THIS venue on THIS surface
  07-20 Kitzbuhel (clay) R32  LOST  Buse 1808         7-5 6-7 6-1   — competitive vs Strong
  07-13 Umag (clay) R32       LOST  Prizmic 1868      1-6 3-6       — blown out
  06-30 Wimbledon R128        LOST  Choinski 1553     3-6 5-7 2-6   — UNDER, straights to a sub-1600
  06-25 Mallorca QF           LOST  Quinn 1593        5-7 7-5 6-3   — UNDER
  SHAPE: FADING season (R1 exits, a 1-6 3-6 blowout), but SHARPENING this week — 6-4 6-4 over 1626 on hard yesterday is his best scoreline in six weeks.

═══ PATH & SHARPNESS (Step 1b) ═══
Kopriva: 1 match at the venue, won it DOM in 90 min — clay-to-hard transition already MADE and validated, conditions-adjusted, timing grooved.
Lehecka: first match at Montreal (direct entry). Last match anywhere 5 days ago — mild rust only, and it was on hard (Los Cabos), so NO surface transition. Rust risk is small; the concern is level, not layoff.

═══ FATIGUE (Step 2) ═══
Kopriva: played YESTERDAY, but a TWO-set 90-min win. Per the calibrated study the -4.5pp hangover applies to 3-set wins only; a 2-set win yesterday is roughly neutral (~-2pp). Age 29 = baseline band, no modulation.
Lehecka: 1 match / 3 sets / 127 min in 14d — fresh.
Verdict: roughly equal, no extreme spot. Fatigue is CONTEXT ONLY and contributes nothing to the edge.

═══ NARRATIVE (Step 3) ═══
Let-down: none (no Final in either window). Venue momentum: Kopriva 1-0 at Montreal this week; Lehecka no window matches here. Title recency: none. Surface transition: Kopriva clay->hard, but already completed and won yesterday, so it is priced-off rather than a live risk.

═══ ELO vs RANKING SANITY ═══
Lehecka: rank 19 but elo_prob 71.9% / sim 65.3% — STRONG RANK, LOW RATING. His numbers are dragged by a losing clay swing (Rome, RG) plus one bad hard result; the rating likely UNDERSTATES a genuine top-20 hard-courter. Framework rule: do NOT back the opponent purely on that gap. This is the main discount applied.
Kopriva: rank 55, but the 2026 body of work (R1 Wimbledon loss to 1553, 1-6 3-6 to Prizmic) looks sub-100 in level — the ranking FLATTERS him.
Both cross-checks point the same way, so I stake at the BOTTOM of the model range: honest Kopriva 25%, not the blend's 26.9% and nowhere near the sim's 34.7%.

═══ MARKET CHOICE (Step 4b) ═══
Betfair ML only. alt_markets.handicap = null, alt_markets.totals = null, alt_markets.model = null — no margin market is quoted, and Pinnacle has no row to supply a sharp games reference, so ML is the only expression available.
AH: not quoted
Edge test at the executable price: 25.0% - 20.83% = +4.2pp -> PROBE band (3-5%). Probe min_odds = 1.03/0.25 = 4.12; taken 5.00 clears it (full-ladder min 1.05/0.25 = 4.20 also cleared, but the 4.2pp edge is short of the 5% full-ladder bar, so this stays a 0.5u probe).
Re-analysis note: the drift from 4.60 to 5.00 improved breakeven from 22.5% to 20.83%, i.e. +1.7pp of edge versus the prior decision. The thesis is unchanged; the price got better.
refutation_check: no web search tool available in this environment and the brief carries no player_notes/web_insights block -> no finding recorded, thesis unchanged (nothing added, nothing removed).

═══ GATES ═══
GATES
  G1 edge-origin: PASS — pre-narrative, models-only: blend B 26.9% vs 20.83% commission-adjusted breakeven = +6.1pp (ELO alone +7.3pp, sim +13.9pp); edge exists from model-vs-market disagreement with ALL fatigue/path/narrative stripped. Staked prob 25% is a DOWNWARD haircut from the ranking cross-check — narrative shrinks, never enlarges. 4.2pp final = probe tier, 0.5u.
  G2 consensus:   PASS — 3 of 3 models (sim, ELO, ML) price Kopriva above the exchange breakeven; full-ladder tier by consensus, but the probe band binds the stake to 0.5u regardless.
  G3 fatal-risk:  PASS — ML bet, G3 cannot block. No self-negating risk: the dog thesis does not depend on margin. Kopriva's poor June/July body of work is the ordinary 75% downside, already priced into a 25% honest estimate, not a contradiction of the edge.

═══ VERDICT ═══
BET — Kopriva ML @ 5.00 Betfair, probe tier 0.5u.
Lehecka at 1.20 is priced as a near-lock (80.6% no-vig) by a market reading rank 19 and a strong grass fortnight. His one hard-court outing in the window is a 6-1 3-6 4-6 loss to a 1701, and he walks in cold to a venue where Kopriva has already banked a 6-4 6-4 win on the surface. Every model disagrees with 80.6%. I discount hard for the rank-19-vs-low-rating signal and for Kopriva's genuinely weak season, which is why this is a 0.5u probe and not a ladder bet — but at 5.00 the price pays for the disagreement.