Oct 8, 2010 0
In the spirit of giving you fair warning, if you’re not into Australian Rules Football or into data analysis, move along before you taint your eyes with the horrible mash-up of the two that follows.
Now, fair warning given, anyone who has had the pleasure of me herding them into an inescapable corner and ranting at them about standard deviation will know that I enjoy playing AFL Dream Team during the football season. There’s nothing quite like hanging over the barrier at a game to yell ‘Oi, ya lazy #^%#! Kick it, don’t handball it!” at one’s star recruit. (Particularly if you’re also yelling ‘TACKLE HIM!!1!’ at your other star recruit who is on the opposing team.) But most of all, I enjoy it because it’s fundamentally a game of statistics, and there are few things I love so hard as I love stats.
And so, I have a bit of a summer project going on this year. The thing with Dream Team is that there are a bunch of players that everyone will have because they’re obviously going to (a) rise in value or (b) be consistent. These players can be picked out quite readily by skimming the media or the plethora of Dream Team blogs and other resources that have come into being over the past few years. The two things that differentiate a great Dream Team player from a middle of the road one are trading strategy and picking up relatively cheap players who unexpectedly come good.
The trading strategy is something I messed up a little this season just gone and will be working on, but my off-season project is all about the latter – trying to determine whether there are any early indicators of players who are about to have a good season. As a first step, I’ve gone through a bunch of data I’ve managed to scrape from the web and hacked together a bit of an Excel model to help me pick out a pool of players to study. (It turns out – not unexpectedly – that there are a lot of players who have a respectable second season after a low-averaging start as a rookie, but very few players who exhibit a dramatic jump in form from middle-of-the-road to Dream Team gun in years two to five. In fact, far less than one would believe, given all the blog and forum chatter around the elusive ‘breakout year’).
Having identified these players, I’m going to look in more detail at their averages, games played, consistency and so forth in the year immediately preceding their ‘breakout year’ to see whether they share any common characteristics not observable in non-breakout players. As a sideline, I’m also going to look at the second-year players who have demonstrated a significant improvement from their rookie form, although I think the reasons for this (and the likely players) tend to be a bit more obvious to begin with. Here’s a screencap of the work-in-process with a bit more detail around the proposed methodology:
Yes, this is truly what I do for fun on my lunch-break. I reckon it beats shopping for shoes by a factor of about eleventy million.