TennisEdge's AI model finds no betting value in D. Galfi vs E. Seidel (Warsaw on 2026-08-03) — the market is efficient, with D. Galfi favoured at 1.91.
Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.
| Method | D. Galfi | E. Seidel |
|---|---|---|
| Elo ratings-based, surface-weighted |
40.9% | 59.1% |
| Machine learning 29-feature gradient-boosted model |
42.2% | 57.8% |
| Serve / return point-by-point simulation |
45.1% | 54.9% |
| Blend what the bot actually prices off |
45.0% | 55.0% |
| Market implied de-vigged from the opening odds |
50.0% | 50.0% |
We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?
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 14, 2026.
Level at 1–1 over 2 meetings. Full record.
| Date | Event | Surface | Winner | Score |
|---|---|---|---|---|
| Aug. 4, 2026 | Warsaw | hard | E. Seidel | 2 - 0 |
| April 2, 2024 | La Bisbal D'Emporda Challenger Women | clay | D. Galfi | 2 - 0 |
| L | E. Seidel | 0 - 2 | Aug. 4, 2026 |
| L | M. Sherif | 2 - 0 | July 14, 2026 |
| L | K. Rinaldo Persson | 0 - 2 | July 7, 2026 |
| L | F. Jorge | 0 - 2 | June 23, 2026 |
| L | S. Lamens | 2 - 0 | June 15, 2026 |
| W | D. Galfi | 0 - 2 | Aug. 4, 2026 |
| L | D. Snigur | 1 - 2 | July 20, 2026 |
| L | A. Kulikova | 0 - 2 | July 7, 2026 |
| L | L. Noskova | 0 - 2 | June 30, 2026 |
| L | S. Sierra | 2 - 0 | June 20, 2026 |
All Aug. 3, 2026 results & picks →
D. Galfi (A, 27, #119) vs E. Seidel (B, 21, #100) — Warsaw, hard.
MODELS vs MARKET
ELO: B 59.1% | ML: B 57.8% | blended model: B 55.0%
Market: both 1.91 (book, ~50/50 with overround; implied B 52.4%)
Value pick B, value_edge ~5% (vs raw implied 50%).
NO exchange price (alt_markets.betfair_ml null) -> book-priced only.
Edge vs BOOK quote for B: ELO 6.7%, ML 5.4%, blended 2.6%.
RECENT FORM (last 5)
Galfi: L 6-2 6-0 Sherif (clay, 1762) BLOWOUT; L 4-6 6-7 Rinaldo Persson (clay);
L 6-7 3-6 Jorge (grass, Wim QF); L 6-3 7-6 Lamens (grass, Berlin SF);
L 6-4 6-4 Mandlik (grass). Won Rome title on clay in May but current shape
is FADING hard -> last out a double-bagel-adjacent 6-2 6-0 loss.
Seidel: L 5-7 6-3 3-6 Snigur (clay, Prague); L 1-6 4-6 Kulikova (HARD, 1445 ELO)
-> poor hard-court result vs weak opp; L 4-6 3-6 Noskova (grass, Wim);
L 6-2 6-1 Sierra (grass, blown out); L 6-3 6-4 Parry (grass). Level-dropper:
ELO/model lean built partly on W75/W50 clay results -> possible farmed rating,
discount. On the actual match surface (hard) her last result was 1-6 4-6.
SHAPE/PATH: Both FADING, both off long losing runs. Both transitioning off clay
to hard. Neither played THIS venue this week. No rust-vs-sharp differentiator.
Galfi ~19d since last match, Seidel ~13d -> no extreme fatigue spot either way.
ELO-fav Seidel ranked 100 but hard-court live form is weak; the model's B lean
is not independently validated by surface-specific evidence.
MARKET CHOICE
alt_markets: none at analysis time
ML only, book-priced. Best-case edge (ELO alone) 6.7%; blended just 2.6%.
Pricing Discipline #4: no exchange ML price -> need edge >= 8% vs book AND
flag book-priced, else PASS. Bar not met. Small event, no exchange liquidity =
the segment the backtest loses on.
GATES
G1 edge-origin: FAIL — book-priced edge 2.6% blended / 6.7% best-model, below
the 8% book-priced bar; no exchange price to price against.
G2 consensus: PASS (tier only) — 3 models lean B but modestly.
G3 fatal-risk: n/a (ML) — Seidel's own hard-court form (1-6 4-6 vs 1445) and
farmed-ELO doubt argue against leaning on the model.
VERDICT: PASS. Model lean on Seidel is real but modest, book-priced only, and
fails the 8% no-exchange bar. Her live hard-court form does not independently
validate the ELO edge (possible farmed rating from W75/W50 clay). No edge to stake.