April 13, 2009

Playoff Push - Does the record from February 1st matter?

Mirtle has a post on how teams have done since February 1st. I thought I'd provide some additional information.

I would argue that the team's recent record is completely irrelevant. What matters most is how good they were for the entire season. The table above is hardly conclusive, I just took all the playoff teams that had a 60% or better record from February 1st on and looked at how they did. The table above is the result of looking at three seasons. You should notice that 2005-2006 didn't have any teams with 60% or better record from February on.

The table shows that these teams Won 2 series and Lost 5 (29%).

1 comment:

Sunny Mehta said...


So funny, I was just thinking about this exact topic.

Let's discard "record" and pick a better handicapping/predictive stat, particularly for smaller samples. It can be whatever stat you like, maybe Expected Goal Differential (with or without adjustments for goaltending), or Actual Goal Differential (with or without adjustments for quality of opponent), etc. Doesn't matter. Some stat that you think has good predictive value.

Now, here's my question. Between looking at every teams' performance in that stat since Feb 1st, or performance in that stat since the beginning of the season, which would be a better predictor of playoff performance?

While I certainly think people generally read too much into "trends" (i.e. - noise over recent samples), hockey teams are not exactly roulette balls. In other words, I gotta think many teams have had better results over the past 30 games because they are better teams than they were in November, not just because they're running good.

For example, you mentioned how the Canucks are kind of a different team now that they have Sundin and a healthy Luongo. It seems kinda silly to factor in Vancouver's results sans Sundin and Luongo.

Or look at the Penguins. According to your expected goal model they are a .500 team over the course of the season. However, they changed coaches, got Crosby and Gonchar back healthy, picked up Guerin, et al, and over the past 20 or so games they're running at like a +15 expected goal diff. Which set of results is closer to their "true" value today?

Just curious about your thoughts.