Repeatability Of One-Goal Victories

One of the downsides of hockey bloggers getting hired by NHL teams this summer is that we've lost a lot of research that's interesting or useful. One thing that I know has been researched that seems to have disappeared is the repeatability of one-goal victories. It's often asserted that a team's record in one-goal games has a huge element of chance to it, and that games decided by two or more goals are a better indicator of how good a team really is. Copper & Blue used to calculate something called Clear Victory Standings based on this idea. But the evidence that they link in order to support the idea is no longer active, and I've been unable to find anyone else who's gone to the trouble of checking.

So I decided to run a check myself. To ensure that the smaller sample size of the lockout-shortened 2012-13 NHL season doesn't lower the reliability of the result, I've gathered data on the three seasons that directly preceded the lockout. It seems unlikely to me that the result would have changed drastically over the past couple of years, so I don't think there's any real question of whether the result will still be reliable.

The rest of this post won't be very long, because what I did was simple, and the result is very clear. I looked at the correlation between 1-goal victories in one year and 1-goal victories the next year, and then I did the same with games that were won by 2 or more goals.

Are One-Goal Victories A Repeatable Talent?

Nope.

There's almost no relationship between winning (or losing) a lot of one goal games in one season and winning (or losing) a lot the next season. There's a reasonably strong relationship between wins by two or more goals from year-to-year, though. It's pretty clear which is an indication of real talent and which is mostly chance.

Can You Put That In A Couple Of Charts For Me?

Sure.