Arlington69 has come up with yet another alleged piece of evidence with relation to momentum. In his latest post, he claims to have evidence supporting the idea that the game invokes deliberate changes in the passing accuracy. We checked it out.
“If your team struggles to pass and keep possession in first half around 90% of the time the second half will show great improvement.”
The above is a quote from the info-graphic accompanying Arlington69’s post dealing with the idea that the game invokes momentum swings in the shape of artificial drops in passing accuracy . The basis of Arlington69’s post is a statistical analysis. As you may have noted, Arlington69 has collected and recorded a massive load of data from his own matces. In this case, he looks at how the passing accuracy changes between the two halves in cases where either player either had a low or a high passing accuracy in the 1st half. The outcome of that comparison is plotted in the chart below. As clearly visible, players who either had low or high passing accuracy in the 1st half usually saw their higher passing accuracy swing in the opposite direction in the 2nd half. Arlington69 refrains himself from claiming that the observations above prove artificial momentum swings. Instead he argues that “What causes it is hard to say for sure but this variability helps to understand why people think the game has momentum“.
Why passing accuracy isn’t constant
There is little doubt that the passing accuracy can vary during a match or between matches, but we are not quite as puzzled as Arlington69 when it comes to why that happens.
The primary factor influencing your passing accuracy is your ability to make the right passing decisions under any given set of circumstances.
A lot of players appear to believe that having a team with top passing stats should guarantee you a high passing accuracy in every match and throughout the match. This is however far from a realistic expectation.
If you constantly make the wrong passing decisions, your passing accuracy will plummet no matter what stats your team has. If you on the other hand make the right decisions, your passing accuracy will skyrocket no matter what team you are using.
Stats will neither save you nor block your path to success. But they do make a difference, and most notably in regards to those passes where the difference between failure and success is razor sharp. It’s when the small margins become decisive that Iniesta presents his credentials. When it comes to bread and butter passing, any bronze midfielder will do just as well.
But risky passes normally make up a small fraction of the passes in a match. The short, easy passes that make up the majority of your passes rarely fail because of stats.
Why does passing accuracy vary between and during matches?
Mainly for the same reasons that human performance varies in all other aspects of life. Among the factors are physical and mental fatigue as well as tactical dispositions of both players. Some opponents prefer to play a game with many risky passes. This creates space for countering if you are willing to play a high risk passing game as well, but this comes at the price of more failed passes.
In regards to variations happening during a match, people sometimes deliberately or inadvertently change their playing style. This may come down to simple things such as lack of concentration, or it may happen because someone deliberately pushes the game higher up the pitch. On top of that, FIFA’s fatigue concept means that players become less likely to move into position and create passing options. Some human players don’t adapt well to this and keep playing as if their players move equally well for the full 90 minutes.
We have presented some of the completely natural reasons why the passing accuracy varies.
Yet, Arlington69’s results most likely have a different but equally harmless explanation: A statistical phenomenon called regression toward the mean. Under the hood of this fancy term lies a phenomenon which all football fans know.
Per January 15th 2018, Paulinho had scored 8 La liga goals, which was exactly twice as many as Cristiano Ronaldo. But when La Liga ended in May, Paulinho had increases his goal tally to just 9 goals. Cristiano Ronaldo had scored 26. Hence, what we ended up with at the end of the day was quite ordinary despite the two players having performed extremely bad and extremely well up untill a certain point earlier in the season. Statistically, both players ended up with scoring tallys which had regressed toward the mean.
The same phenomenon can be found in many aspects of FIFA as well. For instance, a match may have an extremely high or low goal ratio or passing accuracy in a certain period. But chances are that its overall passing accuracy will regress toward the mean of all matches over 90 minutes.
What Arlington69 has done here is essentially to pick out all those matches where the passing accuracy was extreme in the 1st half and compare the passing accuracy in the two halves in those matches. And not surprisingly, he observes that the passing accuracy in most cases regressed toward the mean during the 2nd half.
This is in other words completely normal. In fact it would have been far more sinister if Arlington69 had observed what he apparently expected to see: Namely that the passing accuracy didn’t regress toward the mean.
Arlington69 demonstrates that there are performance swings, which hardly ever was in question. And obviously, that observation alone doesn’t deliver any new answers to why performance swings occur or why some people believe that they are invoked deliberately. So, in respect to generating new knowledge, this study is of absolutely no value.
As much as we appreciate people generating new knowledge about FIFA through statistics, we find Arlington69’s approach to be counterproductive. A lot of people have pointed out problems in his methods. Failure to consider regression toward the mean, inadequate sample sizes and sampling bias are just some of the problems that have been pointed out repeatedly.
Committing the same method errors again and again does not make anyone smarter.
Furthermore, we have seen Arlington69 interchange between different, clearly contradicting conspiracy theories. A couple of days ago, he argued that the game invokes momentum swings if someone takes an early 2-0 lead – but not later in the match. Now, he is testing a theory which involves artificially invoked momentum swings throughout the match. And we have even seen him claim that the game helps out the lesser team, which again is something completely different and completely incompatible with the other claims he has made.
It ought to trigger some reflection and not least reconsideration about the validity of your theories when your own evidence doesn’t fit.