Chelsea v Leeds
How the SharpXI model rated this EFL Cup fixture — every call it made on shots, fouls and tackles, set against what actually happened on the day.
Before Chelsea beat Leeds 6–3, the SharpXI model made 14 confident calls on the player markets and 10 landed — the strongest being Anton Stach for 1+ tackles (94%, had 2). Its clearest miss: Ao Tanaka for 1+ tackles (90%, had 0). Every probability here was published before kick-off, priced the same way as every other fixture — one game is noise, not a verdict.
7 matches on record — too few to quote a card rate against. See the record
Player markets — the model's calls
Each probability is the model's own number for the headline (1+) line, before any bookmaker margin. The result column is what the player actually recorded.
| Player | Market | Line | Model | Actual | Result |
|---|---|---|---|---|---|
| Anton Stach | tackles | 1+ | 94% | 2 | ✓ |
| Sean Longstaff | tackles | 1+ | 94% | 3 | ✓ |
| James Justin | tackles | 1+ | 94% | 7 | ✓ |
| Ao Tanaka | tackles | 1+ | 90% | 0 | ✗ |
| Ethan Ampadu | tackles | 1+ | 90% | 3 | ✓ |
| Daniel James | shots total | 1+ | 89% | 1 | ✓ |
| Valentín Barco | tackles | 1+ | 88% | 7 | ✓ |
| Melvin Bard | tackles | 1+ | 88% | 3 | ✓ |
| Maxence Lacroix | tackles | 1+ | 84% | 1 | ✓ |
| Pep Chavarría | tackles | 1+ | 84% | 3 | ✓ |
| Josh Acheampong | tackles | 1+ | 84% | 3 | ✓ |
| Malo Gusto | tackles | 1+ | 84% | 0 | ✗ |
| Lukas Nmecha | shots total | 1+ | 84% | 0 | ✗ |
| Harry Wilson | shots total | 1+ | 83% | 0 | ✗ |
How these are modelled: Tackles Shots total
Dig deeper
Past meetings
- Chelsea 2–2 Leeds Feb 2026
- Leeds 3–1 Chelsea Dec 2025
- Chelsea 1–0 Leeds Mar 2023
- Leeds 3–0 Chelsea Aug 2022
Frequently asked
How did the SharpXI model do on Chelsea v Leeds?
10 of its 14 most-confident calls landed. The board on this page lists every one with the model's pre-kickoff probability next to what actually happened, hits and misses in the same table.
What did the model get wrong on Chelsea v Leeds?
The most confident call it got wrong was Ao Tanaka at 90% for tackles — shown in the same table as the hits, marked with a cross. A record that hides its misses isn't a record, so we don't.
Are these probabilities hindsight?
No. Each one is the number the model produced from data available before kick-off, using the same settings as every other fixture on the site — nothing was refitted after the result. That is the only way a per-match report means anything.
Does one match tell you if the model is any good in the EFL Cup?
No, and we'd rather say so. A single fixture is a handful of calls and mostly noise; the season-long, market-by-market record — including the markets that fail to beat a naive guess — is on the track record page.
SharpXI models probabilities from public match data; it does not take bets or guarantee outcomes. Prices are entered per leg in the tool. 18+ — please gamble responsibly.