Oilers start is an important reminder that statistics don’t tell the full story

Statistics are a mainstay in sports. After first being used to analyze players in baseball, statistical analysis has also found its way into many sports, most notably hockey, football and basketball.  In fact, #fancystats are now so common in the hockey world that you will rarely read a hockey article that neglects to mention them.

Analyzing hockey players without statistics is like only using one eye.  You can still see, but you don’t get the full picture.  

This can go both ways.  Analyzing a player just with statistics is like only using your other eye.

However, I don’t weight traditional and statistical analysis equally.  I believe that statistics should be used to formulate about 60-70% of one’s opinion of a certain player, leaving 30-40% for traditional analysis.

Not all people share this view with me, which has led to a quite unfortunate view that some people hold, where statistics are weighted at 100%, and are used independently.  Statistics can be an incredible resource, but the should not be used without the company of the “eye test”, or traditional analysis.  Only looking at the statistical aspect can lead to misleading conclusions, such as the conclusion that the Matthews is not the best player on the Toronto Maple Leafs team, which I wrote about in a previous post, which can be found here, or that the Oilers are playing well so far in the 2017-18 season, the subject of this post.

Shot based statistics such as Corsi and Fenwick show that the Oilers are dominating, as they have 59% of all Corsi events and 60% of Fenwick events.  

These statistics tell a completely different story than anybody that has watched the games will tell.  The Oilers have looked disorganized and just plain bad so far, a view that many others share.

Jason Gregor, Oilersnation: “The Oilers aren’t losing due to a lack of talent.  They have enough skill to compete, but not enoug talent to overcome the hideous mistakes they’ve been making.

Bad reads
Ill-advised pinching
A porous penalty kill
Sub-par goaltending
Inability to score from in close.”
(Article)

Jason Gregor is a smart man, and this excerpt sums things up well.  There is a simple message in the pair of paragraphs; the Oilers are making too many costly mistakes. It’s as simple as that.  It doesn’t matter if they are vastly out shooting their opponents, if they can’t cut down on these mistakes, they will not even come close to making the playoffs.

There is no statistic that measures costly mistakes, which is exactly why statistics should not be the only factor taken into account when analyzing a player.  

If statistics cannot account for something that can have such a huge impact on a game as a mistake that leads to a goal, statistics should not be used exclusively to analyze a player.

A common and perfectly valid belief in the #fancystats community is that it is foolish to judge a player solely based on what you see.  This would be perfectly fine, but many statistical analysts exclusively use statistics to evaluate performance, which, in my opinion, is just as bad.

Both allow you to formulate valid opinions about a player’s skill and performance, but when combined, that opinion becomes more informed and more powerful than before.

You would never use just one eye to view something.  Do the same with this.

Remember:

Statistical Analysis = valid opinion

Traditional Analysis = valid opinion

Statistical Analysis + Traditional Analysis = Powerful, informed opinion

Don’t limit yourself to just one eye.

Use both.