Barcelona are ready to enter the race to sign Crystal Palace and England defender Marc Guehi on a free transfer next summer with Liverpool and Real Madrid already interested in the 25-year-old. (Sport - in Spanish), external
Liverpool defender Joe Gomez, 28, attracted interest from both Crystal Palace and Brighton on transfer deadline day but the England international stayed at Anfield as the Premier League champions failed in their bid to sign Guehi. (Football Insider, external)
The futures of English centre-back James Tarkowski, 31, Ukraine full-back Vitalii Mykolenko, 26, and English midfielder James Garner, 24, will come under review at Everton, with the trio having entered the final 12 months of their contracts. (Times - subscription required), external
Manchester United could return for Brighton midfielder Carlos Baleba next year after failing in their attempt to sign the 21-year-old Cameroon international in the summer transfer window. (GiveMeSport, external)
Fulham will offer a much-improved contract to Rodrigo Muniz after they held on to the 24-year-old Brazilian forward despite interest from Serie A side Atalanta in the summer. (GiveMeSport, external)
Emiliano Martinez is expected to be reintegrated into the Aston Villa squad after the 33-year-old Argentina goalkeeper failed to secure a move to Manchester United on deadline day. (Athletic - subscription required, external)
Martinez is prepared to reject offers from the Saudi Pro League and Turkish Super Lig, where the transfer window is still open, in favour of staying at Villa for another season. (Football Insider, external)
Tottenham's bid to sign Jan Paul van Hecke, 25, has been boosted after the Netherlands centre-back decided to hold off from signing a new contract at Brighton. (GiveMeSport, external)
Rangers could return with a new offer for Crystal Palace's Jesurun Rak-Sakyi after the Scottish side missed out with a late bid for the 22-year-old English winger. (Football Insider, external)
Wolves have inserted a clause in the loan deal that allowed forward Sasa Kalajdzic, 28, to join LASK which means they can recall him January. There is also no option for the Austrian club to buy him as part of the agreement. (Express & Star)
The Athletic
What the Lucas Paqueta case tells us about the use – and misuse – of data in football
By Mark Carey
We data folk rarely need a second invitation to discuss statistics and analytics, but when a story falls into your lap, it would be rude to ignore it.
West Ham United’s Lucas Paqueta was cleared of the Football Association’s four spot-fixing charges in July after a hearing that had spanned almost 12 weeks with the independent panel — whose findings were published this week — identifying holes in the two-year investigation, including the use, or rather misuse, of data.
As part of the investigation, the FA leant on analysis of the four suspicious matches provided by Stats Perform Integrity Services (SPIS), which is a division of data company Stats Perform. Their role is to support clients and other stakeholders to identify and take action against match manipulation, particularly those related to match betting.
Using Opta-derived metrics, the analysis looked at how many fouls Paqueta committed, defensive duels he contested and tackles he attempted within each of the games in question — comparing those numbers with his season average to flag any questionable actions relating to the number of yellow cards he received.
On the one hand, it is a good thing that leagues invest in tools that improve the integrity of the sport, with SPIS’s analysis using quantitative and qualitative information to provide a score within an Integrity Grading Index (1-4) for each incident.
However, building a case around a small number of isolated data points — such as tackles and duels contested — was, in all likelihood, always doomed to be flawed and undermines the process of using data in the first place. The analysis itself was criticised during the case by data consultancy MRKT Insights, with examples provided of low-probability events being a common part of football.
In a fluid, low-scoring ‘invasion’ sport, it is very difficult to localise the impact of a single player using data as an objective measure. To infer causality or intent within those actions is even more difficult, if not impossible, when so many other factors are at play beyond Paqueta’s control.
What is the opponent’s style of play? What was the game state at the time? Did West Ham have less possession in that game? More tackles and fouls will naturally occur in games where the team has less of the ball, with more opportunities to make those defensive actions.
The game is predicated on mistakes from players and officials, so much so that the Premier League has a dedicated “Match Officials Mic’d Up” show, with PGMOL chief operating officer Howard Webb tasked with discussing recent incidents from the weekend’s games.
There are further examples within Paqueta’s case, but the principle relates to a wider point on the use of data in football.
Cherry-picking statistics to suit a narrative — a form of confirmation bias, as described in the commission’s view — is dangerous and undermines the diligence that should come from responsible statistical analysis.
There is a duty of care not to use data for the sake of it, but rather to ensure that analysis stands up to proper scrutiny and scientific rigour where possible. We can all use data, but interpreting it in a way that provides meaningful, reliable insight is where you see your biggest benefit.
At The Athletic, we are avid fans of sounding the “small sample size” alarm bell at the best of times — but it is particularly relevant in the early weeks of a new domestic season.
The frequency with which statistics are used within wider football discourse appears to be creeping up, and should be celebrated. The democratisation of data is a good thing.
However, using data to retrofit a narrative or provide a sensationalist depiction of a player or team’s performance after three games is filled with flaws. Sadly, social media rewards these sensationalist takes which are designed to provoke engagement — often for financial reward.
If you look hard enough, you can use isolated statistics as a stick to beat a player with. Equally, you can use others to make them appear as a potential Ballon d’Or winner.
My own career background in academic research required a different degree of rigour before we were able to share our statistical findings and publish a report. Typically, a peer-review process is required where multiple experts in the field provide feedback on the analysis before a publication is possible.
Within that report, the limitations of the analysis need to be acknowledged. Context, caveats and critical thinking must all be considered when outlining the process from start to finish. That is not to say one approach is better or worse than another; rather, even the most scientific approach might still not provide unequivocal conclusions.
The world is messy, it is woolly, it is muddy — sometimes things are not as clear-cut as we would like them to be. In an age where many of us seek instant gratification and “black and white” answers, the truth is that things are often more grey.
With that in mind, you can understand why a selection of fans show disdain towards the use of data in football. When isolated statistics are presented without context in the media, there is every reason to conclude that data does not offer anything valuable to the conversation.
However, that approach is not representative of the advanced work that occurs across the data departments within clubs. Among other roles, there are plenty of incredible data engineers, data scientists and data analysts who use advanced research methods to provide genuine insight into their club’s chances of success.
If a fan’s only window into football analytics is a contextless statistic that flashes on the screen, though, then they might be justified in thinking data was useless. That opinion might change if they saw the sophisticated analysis that exists behind the scenes — even though it is, rightfully, kept confidential to maintain clubs’ competitive advantage.
We are living in an era of data and technology, with artificial intelligence (AI), large language models (e.g. ChatGPT, Google Gemini, Grok) and machine learning becoming buzzwords that are increasingly used in society, even if we do not all fully understand what they mean.
They too have their blindspots — and that is OK. The key is to acknowledge that when forming your conclusion based on the information you have been given.
That brings us back to the analysis in Paqueta’s case. Just because data is used does not mean that reliable, responsible insight is necessarily being delivered.
(Top photo: Jon Hobley/MI News/NurPhoto via Getty Images)