clock menu more-arrow no yes mobile

Filed under:

Let’s talk about stats, baby

Stats explained using Everton examples

Burnley FC v Everton FC - Premier League Photo by Pat Scaasi/MI News/NurPhoto via Getty Images

Okay, first things first, everyone check their calendars and confirm that we are in fact living in 2019. In 2019 there should be absolutely no one who says/suggests/thinks that football stats are unhelpful, useless, or anything less than a mainstay in this sport and how it is understood both by the professionals and by the fans most plugged in with what’s going on.

Let’s imagine a simple scenario. You and your buddy take in a game on the third row, you get to the pub and your friend goes “Richarlison didn’t get a single shot on goal all game” and you respond ‘No wait, remember he had one around thirty minute mark on the counter attack and then in the second half he had a header off a corner that was on target.” You just used statistics. It would be hard to fathom how a person could say this is useless or unhelpful, it’s simply a notation of the facts as they happened in the game.

A statistic doesn’t become less useful simply because it is more complex. Take expected goals for example. Expected goals pools together more shots than any one person could possibly watch and estimates how likely a player is to score from that position (for example, Opta have analysed over 300,000 shots). Models vary, but what it does is it tells you if your striker is taking good shots that just aren’t going in or if he’s on an unsustainable hot streak.

Let’s use an example. Before Everton signed Sandro Ramirez I backed him for success (not proud of it, I have an irrational soft spot for Spanish strikers and it caused me to overlook the facts). If I’d been paying attention I would have known better. Of his 14 goals, five of them were outside the box direct free kicks. In the same window that we bought Wayne Rooney and Gylfi Sigurdsson, he was not going to be taking those kicks for Everton. What’s worse, he had 6.25 non-penalty xG that season from his 92 shots. Over half of those shots came outside the box, and he got a higher percentage of them on target than he’s done at any stop since.

Sandro, Malaga, 2016-2017

My eye test from watching him at Malaga and my personal bias towards Spanish players sent me in a direction the complex stats would have kept me from. Fourteen goals mattered far, far, far less for the future of his career than his unsustainable xG numbers.

Let me give you another example... Gylfi Sigurdsson is currently sixth in the Premier League in expected assists. Now, you can have one of two reactions to this stat. You can either decide that all xA numbers are bunk or you can dig deeper. xA measures the likelihood of an assist on a created chance using data from far more passes than you or eye will ever watch ourselves. What it doesn’t do is distinguish between xA gained from open play and those gained from set pieces. When we look at the chances Gylfi creates, 1.7 key passes per 90 come off set pieces and 1.4 come from open play. Compared this to the guys ahead of him on the xA list, Kevin De Bruyne gets 3.9/4.9 from open play, Riyad Mahrez gets 3.5/3.9, Trent Alexander-Arnold 2.6/3.5, Roberto Firmino gets all 1.5 of his from open play and Teemu Pukki gets all 1.4 of his from open play as well.

So let’s break this down a bit further. Gylfi is clearly getting more from set pieces than anyone on this list. That can be seen one of two ways, depending on how much you value set piece ability. In open play he’s not going to be quite as creative as a lot of players but when the ball is dead he’s one of the most valuable guys on earth. Pukki and Firmino are getting a lot of bang of their buck on their key passes and probably will not continue to do so for the entire season. There are complicating factors. Saying xA doesn’t matter is out of touch, acting as if xA by itself is the end all be all is deceptive. Knowing how the stat works and what factors can inflate or deflate it go a long way in being able to discuss the game in a way that goes beyond our own limited perception during live time in the game.

Of course we could go on about different advanced metrics all day long, I selected the two most common because they are relatively easy to break down. The point is, these stats simply describe what we see on the pitch in a more detailed way than we can keep track of on our owns. There is still interpretation involved, and knowing how the stats are built allow us to have really fun and interesting conversations about what the data in front of us means. But please, don’t be the guy who’s stuck discussing the sport like we’re all still half drunk in a pub in the 80s. The information available to us today is better than it’s ever been and it is an opportunity to know the game better.