Fact check: No shots = momentum swing?
According to a recent Redditpost, the momentum will swing in your favor in the 2nd half if you haven’t had any shots on goal in the 1st half. The conclusion is based on a statistical analysis, but does the numbers really check out?
The claim
Arlington69 has recorded a massive amount of data about his own matches throughout the year. He is also a proponent of the momentum theory. In his latest post, he tests the claim that the momentum will shift in your favor if you don’t have any shots in the 1st half.
This is his conclusion:
“Where I failed to have a shot on target in the first half my team showed considerable improvement in the second half reducing the clear skill gap without having an noticeable effect on the end result.”
The conclusion is based on match stats from 550 matches or more specifically the 24 cases, where Arlington69 went through the 1st half without having a single shot on target. He calculated averages for total shots, shots on target and goals per half for those 24 matches.
This is what he observed:
1st half (total shots / on target / goals) 
2nd half (total shots / on target / goals) 


Opponent average  6.79 4.05 2.04 
5.63 4.04 1.79 
Own average  1.00 0.00 0.00 
4.88 3.46 1.29 
Based on these observation, Arlington69 concludes that he improved in the 2nd half whereas his opponent performed worse. However, Arlington69 still lost 23 of the 24 matches. So, while the skill gap according to Arlington69 was closed, it had no impact on the final result.
Does this prove that the game invokes a momentum swing when you had no shots in the 1st half? The answer is, not surprisingly, no.
Misleading by statistics
Although we have no reason to believe it is done on purpose, Arlington69’s post is another fine example of statistics being used to mislead the casual observer into believing something which isn’t true.
The main problem is that Arlington69 fails to consider whether the things he observe in fact could be normal. If he had done so, he maybe would have realized that shots not necessarily are evenly distributed between the two halves. In some matches, all the shots happen in the same half, and this is of course particularly frequent in matches where the player had few shots in total. If Arlington69 had conducted the same analysis on real football matches, he would have made the exact same observations, even though EA of course doesn’t invoke momentum swings there.
Having a number of matches where all shots happens to fall in the 2nd half for one of the involved players, isn’t abnormal. In fact, the exact observations presented above are most likely normal.
Another look at the data
Arlington69 kindly provided access to his data on an earlier occasion. That gave me the opportunity to run a few tests of my own on a sample which consists of 761 matches. The first thing I did was to check the number of cases where a player had all his shots in one half, but unlike Arlington69, I checked it for the 2nd half as well.
I found 57 matches where either Arlington69 or the opponent took all their shots in the 2nd half.
But here comes the interesting part: I also found 55 matches where one of the players took all his shots on target in the 1st half.
So, what we see is that in roughly 7 % of all matches, either player takes all shots in the 1st half. And not surprisingly, we have another 7 % of matches where either player took all his shots in the 2nd half.
Quite obviously, the fact that 7 % of the matches contained no shots from a specific player in the 2nd half doesn’t prove that the game swings the momentum away from you when you have shots in the 1st half. There are plenty of matches where players had shots in both halves to disprove that idea.
So, why would the fact that 7 % of the matches contained no shots from a specific player in the 1st half lead us to conclude that momentum swings in your favor when you had no shots in the 1st half?
That’s right, it won’t.
Performance variations and skill gaps
Arlington69 concludes that the observations above express “a clear skill gap” between him and his opponents in the 1st half, and that this skill gap was closed in the 2nd half. This statement is perhaps mainly based on the fact that he went from 0 shots on target in the 1st half to 3.46 on average in the 2nd. But it is nevertheless again a misinterpretation.
In reality, it is perfectly normal that the number of shots varies considerably between two halves. In 41 % of all the matches in Arlington69’s sample, there was a difference of minimum 3 between shots on target in the 1st and the 2nd half.
So, when Arlington69 concludes that a clear skill gap between him and his opponents was closed in the 2nd half, he again makes the mistake of assuming that the norm is same performance in both halves. This is quite obviously not the case, and the performance swing he observes here is very normal.
Conclusion
I must admit that I have a hard time seeing the point in invoking momentum swings in the particular situation where someone had 0 shots in the 1st half – but not in a situation where he had 1 shot against the opponent’s 15. So, I must confess that the entire concept discussed in this article appears absurd, and not least when consdering that Arlington69 still lost 23 of 24 the matches where the alleged momentum swing was invoked in his favor.
Neverthess, we can conclude that there is no such momentum swing.