“Marcos Llorente powers Atlético Madrid to big victory at Granada.”
“Marcos Llorente proving key in Atlético Madrid title push.”
“Marcos Llorente: The best player in LaLiga Santander right now?”
“Transformed into scorer, Llorente has Atlético leading Spain.”
These were just some of the headlines that were being floated around by almost every single football news outlet back in February of 2021. Marcos Llorente was on top of the world, and had become a superhuman midfielder that blended incredible physical attributes with a brilliant footballing brain and superb end-product.
But up until his move to Atlético, the midfielder had gone from promising youngster to unhappy benchwarmer at Real Madrid, and seemed destined for a respectable – if unspectacular – career in top-flight Spanish football.
Then came that fateful night on 12th March of 2020 where Los Rojiblancos faced off against Liverpool in a cut-throat CL knockout game. The first leg at the Wanda Metropolitano was a tight affair that was decided by an early Saúl goal, with the Spanish side then taking a 1-0 lead to Anfield. Liverpool went onto dominate the second leg for 90 minutes, and deservedly drew level on aggregate with a customary Wijnaldum header in the 43rd minute. But they failed to find a winner as Jan Oblak helped Atletico weather the storm and bring the tie to extra time.
And that’s when all hell broke loose.
Firmino scored his first goal of the season at Anfield with 94 minutes on the clock to give Liverpool the lead in the tie, sparking raucous celebrations from the players and fans alike. Then out of nowhere, an Adrian mistake allowed Llorente to receive the ball outside the box and smash one in the far corner to subdue Anfield, driving the tie in favour of Atléti owing to the away goals rule. 9 minutes later he scored again, attacking the box, dropping his shoulder left to wrong-foot Jordan Henderson, then nestling it in the same corner to hand his side the lead outright. And with Liverpool in all-out-attack mode, he and Morata combined exquisitely to capitalise on acres of space to produce a 3rd for Atléti in stoppage time. Marcos promptly named his new dog 'Anfield' a few days later, infuriating the Liverpool faithful and making his pet a permanent personal reminder of his exploits that momentous night.
Without context, Llorente’s performance would only seem very good. But the fact that these were just his 7th and 8th goals and 9th assist as a 25-year-old professional footballer made it something special, a coming-of-age match on the biggest stage that vindicated Atléti’s £35M investment in him. This looked to be a turning point, as the goals and assists slowly started to pick up after football’s return from lockdown, and the future looked bright. But what transpired the next season was beyond anyone’s expectation.
Llorente contributed 23 goals and assists in La Liga alone as Atletico pimped Real to the title by two points. His 12 goals were without penalties and his 11 assists were without corners or freekicks. He played everywhere, with Simeone starting him at right-back, right wing-back, right-mid, right-wing, centre defensive-mid, centre-mid, centre attacking-mid, second-striker, and striker over the course of the season.
He had finally established himself as a superstar after years of being consistently unspectacular, and it was just the start. Llorente was in absolutely peak physical condition and in his footballing prime with a team tipped to finally usurp Barca and Real and become Spain’s premiere sporting superpower.
Yet almost exactly a year after all those sensational headlines flooded the news, Llorente looks to have reverted to his pre-Atletico days. His creative numbers are suffering slightly, he is less productive defensively, and his goal threat has gone from elite to appalling.
There are two main reasons that appear to explain this: injuries and his position. Llorente has not had an extended run of full fitness from October onwards, and after that point he seems to have shifted a little deeper on the pitch. Simeone has also yet to utilise him close to the striker, which is where his goalscoring and assisting was at its absolute peak.
But if you look a little closer, a pattern starts to emerge. Llorente’s underlying and expected numbers have dipped a little, but hardly as much as his cold, hard output in front of goal has.
Almost every single underlying creative number has decreased, but by a not too significant number. And the fact that his touches and crosses have increased is in keeping with a change in numbers down to position rather than performance; fullbacks tend to have a lot of the ball and a big hand in getting crosses in, so touches and crosses should be naturally high, whilst their other creative numbers are usually lower than a midfielder.
His underlying goal threat has also not suffered too significantly.
The drop off might seem large, but his xG (Appendix One) per 90 has fallen by 0.10 and his xA (Appendix One) per 90 by 0.06, which is the equivalent of only 3 goals and 2 assists over a full season. Again, this can be put down to position rather than performance.
But then why is he currently 13 goal contributions behind where he was at the same point last season?
The simple answer: he overperformed his expected numbers in 20/21. Massively. His goals, assists, goal-creating actions, shots on target, shot on target %, goals/shot, and goals - xG were all in the 92nd percentile or above. It is simply impossible to sustain this for an extended period of time, and the fact that he managed to do it for an entire year is an incredible feat in and of itself. Just for context, if Lewandowski overperformed his expected stats in the same way Llorente did, he would have ended 20/21 with 84 goals (Appendix Two) in 29 games in the Bundesliga, more than double his already record-breaking 41, and almost a hat-trick a match. Those are the sort of numbers elite clubs produce as a whole, forget individuals.
Llorente scored 12 off of just 4.3 expected goals that year. This means that, over the course of the season, an average finisher would have scored 4 goals from the chances he had generated – instead he scored 8 more. The story is the same for his assist numbers; 11 off of 5.2 expected. The cumulative quality of chances he created was 5.2, which means if his teammates were average players, they would have converted his passes into 5 goals. Instead, they managed to do much better than they should have, scoring 6 more.
So the conclusion is clear; Marcos Llorente did not magically become a midfielder with world-class attacking output overnight. He happened upon an (extremely) extended period of unbelievable finishing, and relied on his teammates to do much the same from his passes. It is also not like he became significantly worse the next season. The drop off in some of his numbers is mostly down to a shift to a more defensive position, and his finishing underwent a foreseeable regression to the mean.
This is not to take away from his 20/21 stats. Llorente might have had a little luck, but the quality of his finishing for that one season was a once-in-a-lifetime event.
Once-in-a-lifetime is not thrown around as a casual term either – there is substantial evidence to show it is true. According to barcanumbers, Messi’s ridiculous 12/13 had him scoring 2.39 goals/expected goal, a one in 1.6 million event. Llorente scored 2.79 goals/expected goal. It is so far off the mean that using its z-score (the number of standard deviations it is from the mean) would be basically pointless because it is such an anomaly. But to humour this path of logic anyways, using Llorente’s finishing as 6.27 standard deviations from the mean it can be calculated that this was a one in 2,769,715,066 event (Appendix Three). The caveat here is that his numbers are of a relatively small sample size. Although it was over the course of the entire season, he only accumulated 4.3 expected goals; not a big number by any stretch of the imagination.
Taking all of this into account, one can definitively say that Marcos Llorente did not really ‘rise’ or ‘fall’. Yes, on face value his goal contribution numbers went from mediocre to astonishing and back to mediocre again, but the actual quality of his performances have continued to be largely the same.
Marcos Llorente is top of the world no more, but he remains an excellent midfielder that blends incredible physical attributes and a brilliant footballing brain without superb end-product.
(one) – For those unfamiliar with xG (expected goals), it basically measures the quality of a shot by viewing thousands of shots from similar locations and in similar scenarios over the past few years and seeing how often they hit the back of the net. If 3 in a 100 similar shots are scored, the xG of that shot will be 0.03. xG per 90 would be the sum of all the xG that came about from each shot during an average 90 minutes for a player, so if a player takes 3 shots with xG values of 0.01, 0.23, and 0.17, his xG for that match will be 0.41.
xA (expected assists) works in much the same manner. It measures the quality of a chance, by viewing thousands of chances that resulted from similar passes in similar scenarios over the past few years and seeing how often that pass directly results in a goal. If 5 in a 100 similar chances from a specific pass were scored, the xA of that pass would be 0.05. xA per 90 would be the sum of all the xA that came about from each pass during an average 90 minutes for a player, so if a player makes 50 passes in a match, 10 of which create chances worthy of a value (passing to your goalkeeper technically has an xA value, but it is negligible since a goal resulting from that pass is a one-in-a-million event), the xA of that match will be the sum of the xA values from each of those 10 passes.
(two) – 84 goals was calculated by using Llorente's overperformance per shot, multiplying that number by Lewandowski's xG per Shot, then multiplying the result by his Total Shots per 90, then finally by the 90s he played in total.
Llorente's overperformance per shot is equal to his (Goals per Shot)/(Expected Goals per Shot), which was (0.24/0.09), or 2.66.
Lewandowski's xG per Shot is equal to his (xG per 90)/(Total Shots per 90), which was (1.16/4.76), or 0.244.
Multiply Llorente's overperformance per shot (2.66 times) by Lewandowski's xG per Shot (0.244), and you get Lewandowski's Goals per Shot if he was finishing at the same quality as Llorente (0.645).
Multiply Lewandowski's new Goals per Shot (0.645) by his Total Shots per 90 (4.76) and the result is his Goals per 90 if the Polish international had Llorente's quality of finishing (3.09).
The final step is to multiply the new Goals per 90 (3.09) by the total 90s Lewandowski played in the Bundesliga that season, which is equal to Minutes Played (2458) divided by 90, or 27.31.
The end result is 84.48, and once you round down, gives you the hypothetical goals Lewandowski would have scored in the Bundesliga in 2020/21 with Llorente's finishing, at 84.
∴ (Lewandowski Minutes Played/90) * ((Llorente Goals per Shot/Llorente xG per Shot) * (Lewandowski xG per 90/Lewandowski Total Shots per 90) * Lewandowski Total Shots per 90) = Lewandowski’s total goals when shooting with Llorente’s efficiency
∴ (2458/90) * ((0.24/0.09)*(1.16/4.76)*4.76) = 84.48
(three) – To convert standard deviations to probabilities, the empirical rule was applied. The Expected Fraction of Population inside Range formula (erf(x/√2) was utilised, where x = 6.27.
The result is 0.999999999638952, and from there to find the event probability simply plug in the result into (1/(1-x), to find an answer of approximately 2,769,715,066.
Therefore Llorente's quality of finishing relative to his Expected Goals over the course of the 2020/21 La Liga season was a one in 2,769,715,066 event.
∴ 1/(1-(erf(6.27/√2)) = 2,769,715,066