Harry Wilson's late winner secured Fulham a 2-1 victory over 10-man Chelsea on Wednesday, lifting Marco Silva's men within three points of the Premier League's top four.

After his appointment on Tuesday, new Blues boss Liam Rosenior watched on from the stands at Craven Cottage, with Calum McFarlane still in interim charge.

Chelsea were dealt an early blow when Marc Cucurella was sent off after 22 minutes for denying an obvious goal-scoring opportunity by dragging down Wilson, with the referee’s decision upheld following a VAR review.

The Blues were then offered a reprieve as Wilson saw a brilliant first-half strike chalked off for an offside call against Raul Jimenez, which was confirmed after another VAR check.

The Mexican made amends after the break, though, breaking the deadlock with a well-placed header off Sander Berge’s cross in the 55th minute.

The lead did not last long, however, as Liam Delap drew Chelsea level just 17 minutes later, reacting quickest to open his Premier League account for the Blues with a timely poke from close range.

Despite their best efforts, it wasn’t to be for Chelsea, as the in-form Wilson produced another moment of magic, controlling Robert Sanchez’s parry off Emile Smith Rowe’s shot before drilling an 81st-minute winner into the bottom corner.

Data Debrief: Wilson rises to the fore

Fulham have now extended their unbeaten Premier League run to six games, and they had Wilson to thank once again.

The Welshman has had a direct hand in 11 Premier League goals this season (seven goals, four assists) – his outright highest total in a single campaign, while equalling his best goalscoring season in the top flight (also seven in 2019-20).

Against Chelsea, he managed a whopping eight shots, three on target, at an individual expected goals (xG) value of 0.78.

Chelsea, meanwhile, have won just one of their last nine Premier League games (D5 L3), while this was their first London derby defeat since a 1-0 loss at Arsenal last March, ending their nine-match unbeaten run in league derbies.