Behind the Numbers: Possession and Winning

I am weird! For a variety of reasons, but one of those reasons is that I enjoy statistics. In fact, I enjoy it so much that I take on the masochistic task of teaching statistics to undergraduate and graduate students. As a result, I have written a lot of articles at 86Forever over the past three years that contain an element of statistics. I have been giving this some thought for a while and a recent discussion with, mainly, Brenton Walters (@CapsOffside) and Joel Prosser (@JoelProsser) has motivated me to finally start a new series examining the statistical numbers of the Vancouver Whitecaps. I want to note that this will never be an attack on any person’s interpretation of statistics. We are all entitled to our opinion and many things are about interpretation. As a result, two people can look at the same data and come away with different conclusions. I will mention this caveat in subsequent articles as well as many of the statistics I will investigate will come from discussions on this forum and on social media (Twitter/Facebook). I have ideas for a few articles that I would like to write, however, I also want to have input from all of you. If there is something that you are curious about and would like me to investigate, let me know in the comments or through our Facebook page or Twitter account.

While a discussion on individual passing accuracy fueled this new series, I think I will save that for a third or fourth article. The reason for that decision is that I feel a different question needs to be answered first. That question is the value of team passing versus possession. During Canada’s disappointing loss to Jamaica, Rituro and I had a quick discussion about possession statistics.

Of course, Rituro’s comment is nothing new for fans of the Vancouver Whitecaps, as we are reminded of this every match when we see the Whitecaps having 30-40% possession.

If possession is not the important stat (and I believe it is not), then it begs the question, what is the important stat to examine. I think the real answer is a combination of all of them, however, the growing consensus is passing accuracy. In this article, we are going to take the first step of the analysis and examine possession. I initially planned to include passing accuracy and shots all in one, but it resulted in way to long of an article, so those will come in subsequent articles.

The first place I went to obtain data on this topic was MLSsoccer.com. Seemed like the best place to go right. Turns out, when you go to team statistics these numbers are not listed (although it is listed in each game’s stats). Time for another source. I decided to go with the trusty WhoScored.com instead. Sure enough, they had the statistics for which I was looking. For this analysis I will use some data from 2017, but also use some historical data (from 2015 and 2016) as those years are complete and thus we can see how the numbers play out over a longer period of time. However, it is worth noting that the numbers this season should not be too skewed as each team has played enough games that there should be some consistency now.

Success is a difficult thing to operationalize. However, for the purpose of this analysis, I am going to define success as winning. Therefore, I am going to use the MLS standings as the measure of success and see how the top teams perform on possession compared to the lower teams. I will present overall numbers, but I will also break it up into home and away so that we can see the differences.

It is no surprise to most Whitecaps fans that the team is last in possession, with 44.4%. For those that feel possession is not a very useful statistic, it will come as no surprise either that just above them in possession are the 5-12-3 DC United (with a -18 goal differential), the 6-11-1 Colorado Rapids, the 5-9-5 New England Revolution, and the 6-8-5 Philadelphia Union. However, what may surprise people is that the 8-7-5 Houston Dynamo sit 19th while the 8-3-7 FC Dallas are 15th.

Of course, statistics can lie, so we need to look into the breakdown of those numbers more. Whether at home or away, the Whitecaps have around the same possession numbers (45.4% at home versus 43.4% away). Most teams have better possession numbers at home, but there are a few that are around equal or have better away possession. Interestingly, Houston, who is 0-7-3 away from home possess more on the road (46.5%) than they do at home (45.3%), where they are 8-0-2. The Whitecaps most recent opponent, the LA Galaxy, have 52.7% possession at home, where they have only one win, but only 46.7% on the road, where they have a league-best five wins.

On the top end, Sporting Kansas City leads the league in possession (56.2%), followed by New York City FC, Atlanta United, Columbus Crew SC, New York Red Bulls, and Seattle Sounders FC. Currently, SKC are first in the West (2nd in points per game) with Seattle sixth (5th in ppg). NYCFC are third in the East, with Atlanta fourth, NYRB fifth, and Columbus seventh. Points leaders Toronto FC and Chicago Fire sit 13th and 7th respectively in possession numbers, with TFC being equal home and away and Chicago having 53.6% at home and 48.6% away. This aligns with Chicago’s record at home (9-0-1) and away (2-3-4) and modestly with TFC, who are undefeated at home, but are second in MLS with four wins away from home and are tied with LA Galaxy for most points on the road (16).

This gives us a picture of the current season, however, what has been the trend over the past few years? To answer this question, I examined the statistics for 2015 and 2016. Initially, I was going to use just the home matches, but felt that maybe possession matters in one context and not in the other, so I examined both home and away, using points earned as the comparison measure.

In 2015, there appeared to be no correlation between possession and points obtained. However, in 2016, that pattern reversed as possession was moderately correlated with points both home and away. This season, the tie does not appear able to be broken, as it doesn’t matter for home matches (0.15) but does for away matches (0.39). I could add in a bunch of statistical significance tests but I feel that might be getting beyond the level people are interested in. Therefore, I will leave those out.

What conclusion can we draw about the importance of possession in Major League Soccer? The answer is none, as it is correlational. However, we can begin to see patterns. It may be that possession is more important when you are away and less important when you are at home. Potentially, as some have suggested, possession is a meaningless stat because no conclusions can be drawn from it. It will be interesting to see if number of passes per match, passing accuracy, and/or shots play a bigger factor in victory...but those will come at a later date!

I will leave you all with one last analysis specific to the Vancouver Whitecaps. The problem with the above analyses is that they are based on an average rather than looking at each match separately. Sorry, I was not going to calculate that for every team in MLS for the last three seasons. However, I was willing to do it for the Whitecaps for this season.

The above table lists each match in MLS this season for the Whitecaps, the outcome, and Vancouver’s final possession number. In victories, the Caps averaged 42.8%. In losses, 44.5%, while in draws, 40.9%. In other words, around the same regardless, but higher in the worse the outcome!

There are many reasons why possession is not seen as a useful statistics in soccer. My biggest issue is that it always adds up to 100%. There are times when neither team is in possession. If I were writing this analysis for a publication reason, I would probably be interested in seeing whether possession is certain parts of the field (or by certain players) makes a difference. I am sure there is something there, but we just do not have the time here, so I will leave that for someone else to accomplish. However, I am curious to hear what all of you think. Anything from this analysis stand out? Anything you feel needs to be included? Come to different conclusion? Let me know!