Spurs Blogsville

War – But Not The Killing Kind

|

WAR – but not the killing kind by BelgianSpur

The last few weeks on VS have been dominated by Transfer Window talk (some will say justifiably so, other can’t wait for it to end). But given that we are peripheral actors, at best, in this window, and that every opinion out there has already been expressed and defended a million times, I thought it might be a breath of fresh air to discuss a topic which is different, although related.

During the transfer window, there is always a lot of talk about player performance. Players already at the club, as well as any potential transfer targets, are put under the microscope and often hotly debated.

Despite what some may say, using fan opinion as a benchmark is about the worst indicator of what a player is actually worth. It’s unreliable, hugely volatile, and rarely objective. This is where stats come into play. Although stats on their own only tell part of the story, stats accompanied with a bit of context are usually a much better indicator than any subjective opinion alone would be.

Other sports, specifically American sports, often place great value on stats. Stats have taken over as the one objective way to evaluate talent, with NFL, MLB, and NBA teams spending millions every year on stats-based analysis. Teams have replaced old-school scouts by math-oriented Harvard graduates, because…it works (for certain sports – more on that below).

For those of you who have seen it, a few years ago, Hollywood made a movie about this revolution: Moneyball (featuring Brad Pitt). The film is based on a book written by Bill James, an American mathematician and sports enthusiast who set out to prove that stats could successfully predict the outcomes of baseball games. His so-called “sabermetrics” were a revolution in the sport.

Other sports then tried to replicate what James had done for baseball, to see if stats-based analysis could apply to them. 20 years on, they have because commonplace in most big American sports, and more to the point, they have been accepted by the general public/fans as reliable indicators.

James’ initial efforts are the precursor to newer, even more refined stats, based on complex statistical models and algorithms. The idea is that those stats are a living thing, and they are consistently being reworked to be even more accurate predictors of success.

The reference stat right now is WAR, or “Wins Above Replacement”. Simply put, the stat first takes a look at what the average performance is, across the league, for a specific position – that’s the benchmark. Every aspect is taken into consideration: offensive contribution, defensive contribution, discipline, appearances, positional importance (not all positions hold the same value, as some see the ball more often…).

Using that “average league performance” as a benchmark, the stat then looks at a specific player, and compares his contributions to the “league average” benchmark for the position.

That delta is then translated into the “wins above replacement” metric, ie what having that player on the field means for the team, and how that translates to winning. The WAR value can be positive, in which case having that player in your team increases your chances of winning games, or negative, in which case you’d be likely better off with another player in there.

Of course, WAR isn’t perfect. There is still a degree of subjectivity in it because people still decide how much weight is given to every input factored in the calculation (although this is refined over time and gets more and more accurate with time). Also, players are in and out of form, so it’s conceivable that the same player could have positive WAR one season, and negative WAR the next (in which case, looking at the average WAR over several seasons can be useful).

It would be interesting to see if concepts such as WAR could apply to football. At this stage, football’s maturity level with stats is still relatively low. Part of it is a refusal to change, and a desire to hang on to historic talent evaluation methods which may or may not be relevant anymore (American sports went through that period as well – it took decades for the change to become widespread).

Perhaps another, more tangible reason is that unlike baseball, basketball and American Football, which are essentially a collection of set plays, football is a free-flowing sport. That makes it harder to break down the game into “phases” with an “outcome”. This being said, certain parts of football, such as set pieces could be considered. Yet even for those plays, stats seem to be very seldomly used.

What is interesting is that WAR is now entering contract discussions. Concepts such as “$ per WAR value” are becoming a thing, essentially trying to define how much “value for money” a player contract represents.

Seeing how the situations of certain players were so hotly debated on VS, despite having very little tangible information one way or another, I can’t help but think that having a common basis for comparison in football would be more than welcome.

Share this article

Because We Love Football