The international break is almost over, and everyone has got coronavirus. Truly an astonishing turn of events. Now we get to find out what football is like in the presence of empty benches, as well as empty stadiums!
In this week’s roundup we have some analytics podcasts for those of you that like the math to go through the ear hole rather than the eye hole, I’ll shine a light on one of the heavyweights of the football analytics world, and we’ll take a look at the research that is using the conditions created by the pandemic to study the effect that fans have on the game, and measure the home advantage.
Football (& Sports) Analytics Podcasts
White dudes with hobbies? You’d better believe there’s a podcast for it! It’s very easy to make analytics incredibly dry (really that’s the default state), but skimming over the hard bits can make it difficult to talk about the genuinely interesting insights that the data has to offer. Here’s a few podcasts that I think have found that balance:
- Measurables – Hosted by Brendan Kent, former Lead Soccer Data Analyst for the Portland Timbers, Measurables hosts discussions with sports analytics professionals from all over. Some very interesting and insightful conversations.
- Expected Value – The Expected Value podcast, hosted by former Lead Soccer Researcher at ESPN, Paul Carr, is another that looks at analytics across all sports, and it takes a similar approach to Measurables.
- Double Pivot — Michael Caley and Mike Goodman discuss the latest goings on in football (primarily but not limited to European football). For the more in-depth analytics discussions, I’d recommend their subscription-level content, which is available on their Patreon.
- StatsBomb — Ted Knutson (CEO) and James Yorke (Head of Analysis) discuss interesting developments in both football and football analytics. They also do a monthly Q&A episode, which is very informative!
StatsBomb – More Than Just a Data Source
Speaking of StatsBomb, their website is also a great resource for analytics content. StatsBomb has received plenty of attention throughout this series so far, and with good reason. In addition to the data on offer, StatsBomb’s website also includes a lot of insightful analysis about football, the data collection process in football analytics, and the methodology behind StatsBomb’s metrics. The kind of content you can find on their website ranges from technical, like this article looking at the latest update to their expected goals model, which has added shot impact height to the model, to the practical, like this analysis of the use of headers in different leagues.
I’m sure this won’t be the last time I mention StatsBomb, but if you are interested in football analytics, there really is no better place to start.
Analyzing the Effect of (the Absence of) Fans on the Home Advantage
The home advantage is well-established in sports, but there is debate about its cause. Some argue that the home team benefits from the effect that fans have on them, their opponents, and the officials, while others attribute the advantage to the negative effects of traveling, or the positive effect that familiar surroundings have on player psychology and spatial awareness in-game.
This discussion has received renewed attention of late, with researchers in the sports analytics world and social scientists in psychology and economics treating empty stadiums as a natural experiment (a real-world situation that isolates a variable of interest in a manner similar to experimental conditions) measuring fan effect.
Same, Same, But Different
There have been a number of articles in sports media, analyzing the fan effect, but the conclusions have differed. The Athletic’s Tom Worville and Michael Cox argued that ghost games serve as proof that there is no fan effect, while articles in the New York Times and ESPN argue the complete opposite. However, I think all three articles suffer from the same sample size and selection bias issues. They were all written before the end of last season and focused on just one league. By narrowly focusing on the Bundesliga before the end of the season, Smith (NYT) and Hamilton (ESPN) end up overstate the fan effects, while Worville and Cox miss the bigger picture by limiting their research to the Premier League.
The most thorough analysis I’ve seen in the media was carried out by 21st Club and The Economist, using more than 1,500 matches played in leagues all over Europe, finding that the home advantage has declined in the absence of fans, and attributing the effect to a decline in referee bias.
If you fancy taking a look for yourself, check out Ben Torvaney’s article demonstrating how to analyze football results using a Dixon-Coles model in R.
The early findings in academia corroborate the findings presented in The Economist. Research carried out by Reade, Schreyer, and Singleton (an updated version of the research discussed in the ESPN article); Bryson et al.; Vincenzo Scoppa; and Nevill, Balmer, and Williams all comes to the same conclusion: the home advantage has diminished, and the referees are to blame!
I’m a little skeptical that post-lockdown games can be treated as a natural experiment, because I think the sudden removal of fans, the stress induced by the pandemic, and the long break from physical activity during lockdown may have impacted player performance in a manner that weakens experimental conditions. However, I think this is a limitation of otherwise solid research, rather than a fundamental flaw (and their findings are in line with previous research).
So it looks like thousands of people screaming obscenities at referees causes them to wilt under social pressure and favor the home team. Perhaps we should be a little more understanding when referees make mistakes, given that it is normal human behavior? Who am I kidding? They’re evil monsters!
Do you buy the findings in the home-field advantage research? Do you think that fans can lift their team, or is it just a case of influencing the referee? Does home-field advantage benefit some more than others?