How Statistics Can Drive Strategy

Given the relative infancy of statistical analysis in hockey, often I will look to baseball to see how statistics are being used to shape strategies both on the field and in the offices of player personnel executives and then analyze whether those strategies can be used in some form in hockey. Of course the fundamental differences between the two sports make doing that challenging. However it isn’t the end result that is always the enjoyable part of a journey; it can be the journey itself.

Let’s look at Game 3 of the ALCS between the Yankees and the Angels for some insight on how statistics influenced decisions made by Yankees field manager Joe Girardi and how that affected the outcome of the game. Up 2 – 0 in the best of 7 series, the Yankees were looking to put another nail into the coffin of the Angels season. A back-and-forth affair finally reached extra innings with the score tied at four.

Girardi had already burned through several relief pitchers, including star closer Mariano Rivera, when he reached back into the bag and brought out young reliever David Robertson to start the 11th inning. Robertson was coming off a very strong campaign (2 – 1 record, 3.30 ERA with a strikeout rate of 13 K per 9 IP) and he summarily disposed of the first two Angels hitters on a ground out to short and a fly ball to left.

At this point, instead of allowing Robertson try to record the final out of the 11th, Girardi went to his bullpen again and brought in Alfredo Aceves to pitch to the Angels Howie Kendrick. Kendrick quickly singled and then scored the winning run on a double to deep left-center field by Jeff Mathis. Hindsight obviously shows that Girardi may have miscalculated by making this move but perhaps the more important question isn’t “what” but “why”; as in “why” did Girardi make this move?

Undoubtedly, Girardi has printouts of individual player’s stats and graphs and charts depicting every iota of tangible match-up information. Included in this plethora of information was a printout articulating the Angels batters’ averages versus certain types of pitches. These charts showed that Kendrick fared much better against fastballs and ironically, Robertson relies heavily on his fast ball to record outs. Aceves meanwhile, is more of an off-speed type of pitcher.

Clearly the Yankees manager made a reasonable decision to replace Robertson with Aceves so what went wrong? Surprisingly, despite Aceves’ success with his off-speed stuff, Kendrick and Mathis were thrown mostly fastballs by Aceves. In fact, both of the balls hit were fastballs. The problem here wasn’t in the strategy employed; it was in the execution of said strategy.

How might the use of data-driven, specific matchups translate into hockey?

Well the closest comparison to the above referenced example might be in how shooters approach a goalie. If data shows that a goalie is particularly vulnerable to shots high to the glove side then an opposing team would be wise to focus on that area of the net.

That type of matchup data would be particularly useful in shootouts where the shooter and goalie are locked up in a true one-on-one battle. The thing that separates baseball from other “team” sports like hockey is that baseball is a series of one-on-one battles between batter and pitcher. Granted, there are eight other defenders (besides the pitcher) on the baseball diamond – and possibly base runners in play as well – but ultimately it comes down to the pitcher making his pitches against a batter trying to hit those pitches and get on base.

In a shootout scenario, the “shooter” is alone on the ice with the goalie in an attempt to beat the goalie one-on-one. If a coach had data which showed a goalie’s weaknesses (i.e. what part of the net his save percentage is lowest) it could conceivably alter whom the coach selects to take his team’s shootout attempts. Conversely, the opposing coach and goaltender might have an idea where the other team’s shooters may try to go with their shots.

This use of statistics might appear to be of relatively minor potential benefit for hockey coaches but when games can sometimes come down to literally the bounce of a puck or a borderline penalty call, playoff spots are won or lost by a single point, any potential advantage should be exploited. Almost certainly coaches have some, if not all, of this data already at their disposal. However, if any coaches are not making use of simple information like this then they are not doing their job preparing their team to win.


Comments are closed.