Phasing Into Analytics: The NHL And SAP Innovate Their Statistical Database


Screen Shot 2015-11-13 at 1.33.28 PM

Phase Three of the National Hockey League and SAP’s analytics revolution was recently released, building on the NHL and SAP’s collaborative partnership. Chris Foster, the NHL’s Director of Digital Business Development, and Eric Blabac, a Data Scientist at SAP, gave SportTechie insights into the latest innovations.


Phase one of the NHL and SAP’s statistical overhaul focused on creating the dramatically different statistical database, and phase two introduced the Match-Up Analysis Tool. Phase three took place this summer, focusing on the back end, which included taking the existing statistical model and exporting it into a SAP HANA database. Foster described SAP HANA’s tremendous processing power: “It has a lightning fast ability to process data in return time that is sub-zero. And because of [that power], there is this new set available for fans.” The latest update to the NHL’s statistical database mainly revolves around customizable filtering with companion visualizations, giving fans the ability to customize and personalize any leaderboard on the NHL’s “Stats” page.

The latest innovations implemented by SAP and the NHL include the ability to look at shot totals by shot types (such as wrist shots, slap shots, and backhand shots) and to analyze the total of goals that were scored by type of shot. Missed shots, shots that hit the post, or shots that hit the cross-bar can also be filtered. Foster explained that, “in a game where inches matter so much, [these statistical filters will] show stories about different players.”

Screen Shot 2015-11-13 at 1.34.46 PM

Additionally, an extensive look at face-offs will be available. “Face-offs are so crucial in each game, in particular at the end of a game,” Foster says, so being able to examine face-off wins and losses, then filter based on what zone (offensive, defensive, or neutral zone) the face-off took place in, and even further specify by what the game situation was by examining score of the game at the time of the face-off will determine the “clutch factor”.

Screen Shot 2015-11-13 at 1.35.25 PM

New player attributes were also created as filters, such as birth city, state, and province—“making it possible to search the goal leaders of players from Wisconsin,” Foster said while explaining the latest features. Other filters will include draft year, draft number, or draft rounds. Goaltender statistics will evolve as well—allowing fans to study a goaltender’s productivity after a certain number of days’ rest (or a full team’s productivity after a certain number of days between games).

Blabac explained how SAP created a team power index as well, to adjust the standings. This will take a team’s record and point percentage to create a balanced look at the current standings based on their actual productivity during a game.

Previously, the database was not updated until, at the earliest, the post-game. The latency period has shrunk to the quickest speed thus far—approximately thirty minutes, or one period of hockey. In order to collect the data for the databases, there is a real-time scoring feed from each arena. There are six scorers to focus on various events on the ice. Through a customizable solution, the scoring detail and ice events are put into a master feed, which is then sent into the database. And finally, through specific filters and calculations of that data, fans can dissect and organize the information anyway they want.

Screen Shot 2015-11-13 at 1.37.59 PM

Prior to their partnership with SAP, the NHL’s statistics had not changed much since the early 2000s. But since collaborating with SAP, the NHL has seen the processing power of their services multiply, which has allowed the NHL to transform their statistical databases. And the NHL looked to hockey writers and bloggers, the analytics community, and fans to understand where the demand was for the NHL’s statistics. Foster says the NHL has been working with SAP for about six months to create their database, and went through approximately six to eight iterations before reaching the final product. According to Foster, actually creating the database was the most difficult part of the process, and now adding filters and views as they go is much simpler.

From the NHL’s perspective, creating such an in-depth database was intended for fans to engage deeper in the process and analyze different aspects of the game. The database gives fans the power to do any searches or answer any queries. At the same time, phase one brought in new fans. These analytics, according to Foster, are not just for the hardcore fans, but to pique the interest of the more casual fan. The expansive database opens up new types of analysis to better tell the stories of specific players and teams. And the NHL has looked to fans for feedback on their statistical progress. Along with receiving feedback from fans, NHL clubs and front offices have contributed their assessments. The tools provided by the NHL not only provide statistics to fans, but to broadcasters and to front offices. Foster says, “it will be interesting to see what type of analysis is distilled out there. [The database] allows everyone to have their own insights presented in a way that wasn’t before.”

Thus far, the NHL and SAP’s partnership has proven to be successful. Phase one provided the NHL with improved metrics of visitation on their website. It was noted that fans were spending more time on their website, particularly on the statistics section. Phase two saw the creation of the Match-Up Analysis Tool for NHL’s Bracket Challenge, which was a more popular initiative. Foster said the NHL saw their highest record of users, “with just south of one million” using the tool. Plans for the next year include “blowing it up more and making it more robust” for fans. “The consensus from fans has been great. We ask, what are new features that fans want? Over the last six months, [the NHL and SAP] have been building different views and filters and [the plan is] to roll in more phases as the seasons go,” Foster said.

Foster stressed how the NHL hockey analytics community has been around for at least ten years, and by no means does the NHL claim to have created the analytics themselves. Rather, they are trying to build off of the trailblazers in the community. But in their analytics revolution, they looked to ensure that the data is one-hundred percent authentic and actually made by the NHL.

Since Foster and the NHL cannot ensure that all thirty-party websites are completely authentic, the solution was to record the statistics themselves. And by endorsing analytics and creating their own platform, the NHL is looking to bridge the gap between the hockey analytics community and the fanbase. According to Foster, the league platforms allows for a large reach to introduce even the casual fans to more enhanced statistics.

Foster attributes the extensive filtering and ability to dissect data in such a customizable way to differences between the statistics collected by the NHL and other professional leagues. Furthering that notion, Foster said that the SAP and NHL’s collaborative database is “more advanced and more robust than what SAP has done with other leagues.”

Blabac added that “the lowest granulator for most sports is the game-player level, but in the NHL it is possible to go to every single event [from a single game, as well as analyze] every event a player has done in the past five years.” With NHL statistics, there is a wide range of attributes that can be presented.

Questions have arisen regarding the efficacy of utilizing advanced statistics when analyzing NHL teams. Foster said that the NHL is simply putting the toolsets out there for anyone to analyze, but these statistics are a key predictor for when a team has the puck on the ice. These statistics are just a proxy though, and it still depends on a team’s stance on analytics and how to—if at all—analyze the data. “What we want is to make sure that the tools are out there, so anyone—fan or broadcaster has the right tools. The question then is, what do you do with them?” said Foster.

So far, the NHL has not actually created their own statistics, but focused on the different contexts for each of the existing statistics. For now, the context of the data is the focus, but “creating new statistics is on the road map. Once actual player and puck tracking is implemented, the NHL, SAP, and newest partner MLB Advanced Media, will look at creating new statistics such as completed pass percentage and true puck possession.” When that data becomes available, the focus will then shift to creating new statistics. Until that point, the NHL is just looking to advance the data they already have.

Phase three unleashed a new design, look, and feel on the already existing statistical database with robust and advanced filtering. Now, attention is being focused on the next phase, which is set in honor of the NHL’s Centennial anniversary. The NHL possess every game sheet since 1917 and is working to digitize those statistics. Then, customizable features will be added—allowing fans to filter through the entire record of the National Hockey League.

Expanding the database to include every single statistic from the National Hockey League will allow fans to explore at a level that has never been seen before, which Foster says will be “a gift for the fanbase.”