A Quick Note on Shutouts

We here at Daily Faceoff are against journalistic ignorance (even though our loyal readers may think we are in that boat at times), and Hawerchuk at Behind The Net has brought to attention a great example.  Rory Boylen at The Hockey News posted an article in which he pointed out what he thought to be some “surprising” statistics.  Unfortunately, most of these statistics were subject to incredibly small sample sizes or significant biases.  The one that caught my attention was how he noted that Washington and San Jose, last season’s 1st and 4th most prolific offences, had been shutout seven times apiece this season.  Hawerchuk commented simply with small sample sizes, which is perfectly accurate, but I wanted to dig a bit deeper.

So far this season, NHL teams have been shutout in 7.1% of their games, ranging from 0% for St. Louis to 15.1% for Toronto.  Assuming a normal distribution (which may not be perfect but is good enough for the time dedicated to this exercise), we would expect 95% of teams to be shutout between 0.001% and 14.1% of the time.  Curious as to where Washington and San Jose fall?  13.0% and 13.2%, respectively.  At the edge of of the range for sure, but not outside the realm of statistical probability.  In fact, only the aforementioned Blues and Leafs fall outside two standard deviations, which makes perfect sense.

We are drawing on 1600 total games of data here, so I am fairly confident that there is some significance to these numbers, and they would stand up to a full 82 game schedule. Taking it a step further would involve setting up a simulation that takes into account the rate at which teams score, but that is too time consuming for the minor point I am trying to make.  For the record, San Jose and Washington are middle of the pack for scoring goals, so in those specific cases, I doubt it would push them outside the realm of expected results.

I’m not saying it isn’t fun to quote statistics like this, or that we need a hockey equivalent of Fire Joe Morgan, but journalists should at least acknowledge the shortcomings of the data in their articles.  It’s the least they (and I use they, because I am not a journalist) can do.

As a final point, one interesting thing I discovered is that Eastern Conference teams were shut out 66 times, as opposed to the West at 48.  That is a fairly sizable difference.  Overall goals for is pretty even between the conferences (2290 to 2217 for the West), so I am wondering if Western domination in inter-conference games is the cause.  If anyone wants to take the time to look that up, I would be very interested.