So the crux of that Xavi article on whoscored.com (link) was determining a players value to his team through the use of a ‘with or without you’ (WOWY) analysis. As I’ve mentioned before this is something that has been kicking round on the blog of the superb sports statistician tangotiger since about 2006 (link). I think, used properly, WOWY has massive potential for teasing individual performances out from that of a team and examples of this have begun to become widespread through the hockey blogging community.
WOWY analysis was relatively recently adopted by hockey bloggers and one of the earliest, and best, pieces I saw was this post (link) at mc79hockey (link). Mc79, or Tyler Dellow, is a seriously talented writer who’s blog has been around since 2006. In particular his analysis of the goaltender market in the NHL is sublime. I believe he was amongst the earliest to recognise that the gap between the best and worst NHL goaltenders really isn’t much and, in a sport with a salary cap, it really doesn’t make sense to allocate your $ to that position.
Anyway this article takes a player, Alexander Ovechkin, and then compare how his team-mates perform both with him and without. As of today this piece of analysis is relatively easy to do, thanks to awesome timeonice.com (link). It used to take a while to learn the source codes but an html front page has since been developed (link), which can scrape the relevant data from the game logs posted on NHL.com and give you a report with the relevant numbers. It’s truly remarkable how easy it is to access NHL data by the way.
*Required knowledge: Corsi = goals + saves + wide shots + blocked shots
Based on that post and others of the same ilk I’d like to propose a couple of rules for a basic WOWY analysis, applicable across any sport:
We want to judge players based on an event that occurs a significant number of times each game (to reduce random noise)
The event that we choose should be one that is indicative of driving the play in the right direction (so we know whether said player has a positive or negative effect upon the team)
The stat used in hockey? Shots, in this case Corsi as defined above. For an individual player there can be 2,000+ of these events per season. In football we’re looking at closer to 1,000 shots for a team but that is still a number that is far more statistically significant than using results.