Skip to main content
news

SAP statistical analysis explained

Data scientists determine who has the advantage

NHL.com @NHL

What goes into winning the Stanley Cup?

There are obvious things like a deep offense, a smart defense and a goaltender able to make all the important saves.

But is there a way to quantify those attributes? The data scientists at SAP will attempt to do just that during the 2016 Stanley Cup Playoffs.

Using an intricate algorithm that measures 37 different variables, SAP will determine which team has the statistical advantage in each postseason game and series.

Among the variables in the model SAP created are 10 easily identifiable statistics compiled during the regular season, like goals-against per game, penalty kill success rate, points percentage, and shots-for and shots-against per game.

Those 10 factors will be displayed on every series page for the duration of the Stanley Cup Playoffs, but they won't tell the whole story. For example, one team might be ahead in eight of the 10 displayed factors but the other team could have the overall advantage; that's because there are 27 other factors not being shown -- among them things as varied as teams' success in recent games to how much time they spend in various game situations.

There also a few advanced statistics that are part of the model, among them is Shooting Efficiency, which measures the percentage of all shot attempts that become shots on goal.

Another factor is save percentage on high-percentage shots, which attempts to quantify how well a goalie performs against the most difficult shots. Imagine a trapezoid going from the faceoff dots down to the goal; save percentage on shots attempts from that area are considered a big factor in winning games.

SAP also will use a similar analysis tool as part of the Stanley Cup Playoffs Bracket Challenge. "Match-up Analysis with Insights from SAP" will be utilized for every hypothetical series to give fans extra insight when making their picks.

The model used for Bracket Challenge varies slightly from the one used for each game and Playoff series, as some of 37 variables do not apply when looking at hypothetical future-round matchups.

View More