NHL Overtime Game Conspiracy - Update

Back in the first month of the NHL season, I noticed that games played between teams from different conferences had a much larger chance of going to overtime. It was still early days though, and the small sample size (94 games) had something to do with those surprising results (50% of non-conference games went to OT compared to only 15% of matches between conference foes). Now that we are at the all-star break and more than half the season is in the books, I thought it would be a good time to update the results.

Out of 735 total games played through January 25th, 164 (22.3%) have gone to extra time. Of the 554 conference tilts, 112 (20.2%) went beyond regulation while games featuring East vs West have an OT rate of 52/181 (28.7%). The numbers have obviously regressed towards the mean, which is what is bound to happen when you move from a small sample size to a larger one, particularly when the initial results are so skewed. Although there is a noticeable difference between the two, it is still not enough to cry foul. More data is required to prove something meaningful.

I looked at the results 2010-11 and found them to be quite similar. Including last season's 1230 games along with games so far this season, a more statistically significant result emerges:

       Total Intra Inter

Games  1965  1514   451
OT      461   334   127
%      23.5  22.1  28.2

The difference between 22.1 and 28.2 may not seem like a lot, but the it is actually 27.6% (that's 28.2/22.1). In other words, inter-conference games are 27.6% more likely to result in 3 points, a statistic worth considering when gambling in Las Vegas, if you could bet on this sort of thing.

Still, a season and a half is not much at all, so I continued to include prior seasons, planning to go back until 2005-06, which is when the shootout was instituted. However, from that season until 2007-08, clubs played only 10 non-conference games compared to 18 in later years. I decided to ignore these seasons as the difference in scheduling might bias the outcome. (As an aside, they didn't change the results much, mainly because there were so few non-conference games).

I was therefore left with 3.5 seasons to analyze. When I add 08-09 and 09-10 to the data, the theory vanishes, as you can see in the following table:

       Total Intra Inter

Games  4425  3434   991
OT     1044   798   246
%      23.6  23.2  24.8

So is the recent increase merely statistical noise or a changing trend? Let's look at the data by year. The following shows the percentage of overtime games by season.

Season  Total Intra Inter

11-12   22.3% 20.2% 28.7%
10-11   24.1% 23.1% 27.8%
09-10   24.5% 24.6% 24.1%
08-09   22.9% 23.8% 20.0%

As you can see, overtime is becoming more likely in non-conference games every season, while the other numbers bounce around a bit with no discernible pattern.

So do we have a conspiracy? Are teams playing to guarantee each the single point when East faces West? Doubtful. If they were, I think we'd see far more overtime games in these cases. As I said before, I think that clubs might change their late-game strategy when playing a close non-conference game, since allowing their opponent an extra point is not harmful in the conference playoff race.

Then again, we saw the opposite in 08-09. Naturally, such a limited data set makes any truly meaningful conclusion difficult; but the trend bears watching. I'll be paying close attention to the end of game action when I am watching the Leafs in Alberta next month, just to see if there is any noticeable change in strategy.

Best,

Sean