With the NHL postseason in full swing, I’ve started seeing pieces floating around the web suggesting that possession measures can predict the outcomes of playoff series. If you’ve kept up with what I’ve been writing recently, you can probably guess that I’m somewhat skeptical of this idea: game-to-game puck possession might be more consistent than shooting or save percentages, but it’s still very volatile. Single games and short series in the NHL are very random, and specific predictions can be very difficult to make. So I decided to dig into the past seven seasons’ data to examine the use of Score-Adjusted Fenwick to predict the postseason.
Step one of this involved pulling data from stats.hockeyanalysis.com and constructing full-season estimates of Score-Adjusted Fenwick (SAF) for all NHL regular seasons from 2007-08 through 2013-14. Unfortunately, postseason series-level Fenwick/Corsi data are only available back to the 2012 playoffs. Rather than turn this into an enormous coding endeavor, I used a shortcut to estimate series possession. In the course of extensive historical research on NHL seasons prior to 2007, Benjamin Wendorf has used teams’ shots for and against in the first and second periods as a proxy for possession; I grabbed these from the game boxscores and constructed possession stats for all 102 completed series in this way. (As I write this, the 2014 Conference Finals are still undecided.) I used the even-strength goaltending information from the same boxscores to estimate series PDOs as well.
One thing to note right away is that you shouldn’t expect a team’s SAF in the regular season to be reflected in the postseason, for the simple reason that the competition is better. The Kings haven’t had gaudy Fenwick numbers in the playoffs this year, but they’re not playing teams like Buffalo or Edmonton anymore either. Of the 204 team-seasons involved in the 102 series in my dataset, teams underperformed their SAF in a playoff series 131 times. To get at the true expected possession differential in the series, I calculated a simple adjusted % for each team as (Team A SAF)/(Team A SAF + Team B SAF). This adjustment helps a bit, but there’s an obvious problem: the tougher the match-up, the less useful possession numbers are for differentiating teams. The idea that SAF predicts playoff series takes for granted that good regular-season possession teams play strong possession hockey in the postseason, and this is what drives their success. Yet only 64 of the teams in these series expected to control possession actually did so in the series they played. Granted, 45 of those 64 teams won their series, but 19 of the 38 teams that got outshot despite an expected advantage managed to win their series also. More generally, 40 teams won series despite underperforming their expected possession advantage.
So, why does SAF appear to be such a robust predictor of series outcomes? Part of the answer is almost certainly luck: 77 of these series were won by the team with a higher series PDO. If you bring together puck possession and PDO, of course, you’re nearly unbeatable: of the 51 teams who controlled PDO and brought an expected possession advantage to the series, 45 were winners; of the 53 teams that carried both in-series possession and PDO, 46 won. Possession matters, but the PDO effect is unsurprisingly more powerful. Another important piece to the puzzle appears to be home-ice advantage. No one disputes that teams with stronger regular-season possession numbers tend to win more, which implies that they’re more likely to have home ice in the playoffs: this brings them a real advantage in short series. In these 102 series, home teams won almost 57% of games, and the advantage didn’t differ depending on which team entered the series with a higher SAF. Yet through seven playoff years, teams with a higher regular-season SAF have played 57 more games on home ice than teams with lower possession numbers, and have 31 more home wins. The implication is obvious: home ice does more for you if you’re a better possession team in the regular season.
Obviously, we’re dealing with a sample of 102 series, and the results will tend to be driven heavily by (a) more one-sided first-round match-ups and (b) the play of 14 Cup Finalists. These limitations aside, the best way for analysts to maintain their credibility is to avoid making bad predictions, and the unpredictability of hockey outcomes in short series makes them a poor fit for shot-based metrics better suited to bigger-picture research. Teams with better possession numbers will indeed tend to win more often in the playoffs, but this doesn’t mean that teams will necessarily play the way those numbers imply, or that they’ll be successful doing so. Possession matters in postseason hockey, but it’s hardly the beginning and end of meaningful analysis. In a short series, it’s still much, much better to be lucky than good.