When you watch a game, whether it’s on TV or in the stands, whether you follow the match via text, twitter, some fancy JavaScript GameCast, a liveblog or listen to the match on the radio you are, in essence, gathering data. You probably don’t think of it that way but that’s what you’re doing. All of us, collectively gathering loosely defined data on an (or set of) experiential event(s).

Some of us then use the data we gather to tell a story. Maybe it’s a match preview (like the excellent one here from Arsenal Letters) or maybe it’s an opinion piece like Arseblog does every day or maybe it’s a “by the numbers” piece like I do after the match. Or sometimes it’s a piece whereby I look at some hidden data that you might not have thought about or might not have noticed.

You can’t notice everything, sorry. Because what data you gather is dependent upon your filter. No one on this earth can possibly watch a football match and gather every piece of data. It just can’t be done. Hell, I challenge you to set a single event you want to record, well defined in advance, and just gather that information. Then ask your friend to watch the same match and gather the same information. I bet you that you don’t agree on the numbers and the reason for that is that each of you has a filter.

Some people are aware of their filters, most are not. For example, when I watch Gervinho play football I want to throw things at the TV. Those of you who have been reading my blog already know that I have a hard spot in my heart for the Ivorian. It’s totally unfair, I know. It’s totally irrational, I know. But it’s there and I’m aware of it and I am freely admitting it.

But being aware of my filter, I can set it aside and watch him and write about him somewhat objectively. But to get truly objective, I need other people. I need my readers to call me out. I need to talk to other Arsenal fans. I need data sites like WhoScored and Transfermarkt to help me remember certain events and to record those events so that I can retrieve them easily. In short, I need to see the games from other people’s perspective.

Which is exactly the service that Opta provides. Opta has teams of observers watching every game and recording a huge set of pre-defined events. By having multiple people watch the same game and record events that are well defined these stats gathering sites are attempting to bypass our filters. And I think that bothers some people, greatly.

Turns out that some people don’t like to have their preconceived notions challenged. I do. I have no problem with it. For the most part, I like Opta’s definitions and I understand the way they record their data. So, it makes sense to me to use their data, along with other people’s experiential data (opinions, etc) and my own experiences of the matches to tell a story.

Note, I specified “a story” not “the story” or “the whole story”. It’s strange to me why people keep saying that. “Stats don’t tell the whole story.” Of course they don’t. The only people I know who claim that stats are intended to tell the whole story are people who like to mock the use of stats.

Passes completed, for example, doesn’t tell the whole story of passes by a player and I can understand why some people get frustrated when they see something like “Arteta has completed 92.5% of his passes this season”. All the pass completion rate tells us is that Arteta is unusually tidy with the ball. Best in the Premier League, in fact. But does that make the stat useless? NO! I just told you that he is the best passer in the League. If I wrap that in some other data, such as the fact that he has the lowest turnover rate among Arsenal midfielders, that he completes more passes than any other, you start to get a sense of what his game is like. Not the entirety of his game, but the part that I want to show you, right here, right now, for the purpose of this article.

If I want to describe what types of passes he makes, in order to tell a different story, I would need finer granularity. That’s why there are various categories for passes, which Opta does record. They currently record passes forward, passes backward, square balls, backpasses, passes in the final third, throw-ins, set pieces, long balls, through balls, key passes, assists, and so much more.

Here’s an example of the screen shots for all the passing data available via the 442 app (this is Opta’s data).

Not every stat is recorded and not every stat is recorded properly but as you can see there is a huge amount of data out there and it’s all useful.

This idea that not every stat is recorded is actually the critique that I keep hearing in the comments and on twitter. For example, the imaginary stat that I keep seeing make the rounds right now that everyone trots out whenever they want to “disprove” the importance of stats is the “passes he should have made” stat. Even if you could get 10 people to agree to a definition of that stat, how is the number meaningful? You don’t know if the “pass Ramsey should have made” would have resulted in a shot, or if it would have been intercepted, or punted out of play, or any number of other actions that we could all imagine should have or could have happened.

That’s not to say that I agree with the way that every stat is counted or that every stat is infallible. No one would say that. One of the most dubious of stats is the assist. It’s dubious because basically the penultimate person to touch the ball in the build up to a goal gets an assist. It doesn’t matter if Walcott is trying to dribble and loses control of the ball and Santi strikes it home, that’s counted as an assist. And it’s counted that way because if you have to try to judge intent, you could spend hours arguing that. Giroud’s assist to Nacho in the Swansea game is another example. Does he tap it to Nacho or does he lose control? I’ve watched it a ton of times and I still can’t decide. So, it’s easier if they just award an assist for the final pass whether intended or not.

You probably use stats all the time. The League table is a collection of stats. And we Arsenal fans all want to be top of the League table again some day, right? We want to have the best team, as measured by a system in which 20 teams play each other one home and one away game each, which awards three points for a win, one point for a draw, and zero points for a loss, which takes into account goal difference in the event of two (or more!) teams finishing equal on points, and which has a goals scored second tie-breaker.

Those are stats, folks. And you use them all the time to make your point, sometimes ironically about how much you hate stats. You don’t hate stats. You love stats. What you really hate is the point someone’s making with stats.

Qq