After being inundated with ads last weekend, I took a closer look at the Audi Performance Index. I have no doubt that the analysis team behind this metric is collecting a ton of information, but distilling it to a single number doesn't tell us anything new. Here, I offer up some suggestions for making the data more relevant: https://andthenthehex.wordpress.com/2016/03/17/ten-suggestions-for-the-audi-performance-index/ Ultimately, I think there's breakthrough potential for position tracking data. Exploring just how players manipulate space on the field would greatly enhance my understanding of the game.
I think the Audi player Index is cool but I don't think soccer can be entirely quantified by stats so it will always be limited. One thing I will say that they should do is to change the scoring system. For example in baseball if Bryce Harper has a WAR of 10 at the end of the season it is very easy for the casual observer to understand that he provides his team roughly 10 wins compared to a replacement level player. Dom Dwyer having an Audi score of 1600 means nothing to me. I'd like for the Audi Index to tell me how a player does compared to other people at his position and how many points he helps his team earn over a replacement level player. However, like I said before this will never be perfect in soccer because it is much more difficult to quantify than baseball is.
Offense in baseball is much easier to quantify because the game is so individualized with each player having his own at bats. The defensive metrics in baseball are not very good because it is much more difficult to quantify defense. Soccer doesn't have the equivalent to an at bat in baseball so I don't think the Audi Player Index could ever be as good of a measure as WAR is in baseball (and even WAR is not exactly perfect).
Basketball metrics are ridiculously good--perhaps even better than baseball. However, soccer isn't there yet. It's more rigorous than any sport was a couple decades ago.
Statistics will undoubtedly be able to be applied to soccer. The issue right now is that we aren't good enough at data collecting to be able to apply them appropriately. One thing to remember is that statistical anomalies happen all the time. A single extraordinarily rare event may not be unexpected over a large enough sample size. Further, there will likely be "unquantifiable players" around for quite some time. Players who undoubtedly make their teams better, but our statistical models currently have no way of accounting for. Over the last 40 years, that gap has been almost completely closed in baseball. That gap is closing in soccer, but expect it to take some time. There are a couple things which will help statistical analysis of soccer. Ubiquitous television coverage. There are so many leagues with so many players that have every game televised, that we have a huge database of well-recorded data to draw from. That pool of data is only increasing over time. Improvements in statistical gathering. Other sports which are much more fluid than baseball, like hockey and basketball, are getting a whole lot better at quantifying a non-discrete game. Taking lessons learned here and applying it to soccer will be a huge benefit. Lots of money, but also lots of places for that money to come from. A league like MLS which has a salary cap, or a smaller owner who has a more limited budget, can be a great place for statistical analysis to mature. It wasn't the Yankees, Mets, or Cubs that started seriously applying statistical analysis to baseball at a professional level. It was the smaller team (Oakland) that had to compete with the big boys. Data mining. Our ability to tease slight influences out of huge data sets, primarily used for advertising today, can be applied to the large amount of data needed to quantify a soccer game. These things are currently quite expensive, but as we get better and better at it, it becomes cheaper and cheap. Further, as statistical analysis starts making more and more waves in soccer, more and more money will be spent on it. With anything that has as much money flying around as soccer does, statistical analysis will inevitably sneak its way in, and get more and more influential over time. Yes, some teams can just throw money at the problem, but there will always be a team that has far fewer resources trying to figure out how to spend smarter to win out the day.
Good post, and I thnk #2 and #4 will be especially beneficial. I think the next breakthrough in soccer analytics will come from spatial understanding, particularly in relation to teammates and defenders. To use an example from another sport, one of the things most impressive about Steph Curry's insane season is how effective he is regardless of the defensive pressure. A deeper understanding of how players find or create space would revolutionize player scouting and identification.
http://matchcenter.mlssoccer.com/ma...chicago-fire-vs-new-york-red-bulls/audi-index is for last night's Red Bulls at Chicago game. If you click on a player, at the bottom it says: "Computation of the Audi Player Index score is proprietary information and as such, some scoring componnets are not listed here." If the purpose of the Audi Player Index is to entertain fans, make money for Audi, and give TV coverage the ability to compare two players, it works. However, would fans who watched games and thought something about a player believe a proprietary formula provides a better evaluation than the fans' own eyes? If the goal of the Audi Player Index is to convince fans to evaluate players differently, that can't happen without telling the public what the formula is. To moderators: I didn't know if I should start a new topic about this or bump up this topic from March. If bumping up this topic was wrong, please use this post to create a new topic titled "Why Should We Believe The Proprietary Audi Player Index?" The new topic should include all of my post until this "To Moderators" section. Please do not lock this topic without using my post to create a new topic.