In a continuing effort to phase the National Hockey League (NHL) into an analytic future, the NHL and SAP HANA Enterprise Cloud service have created the Match-Up Analysis with Insights from SAP. The Match-Up Analysis tool was created to help fans compete in the NHL Stanley Cup Playoffs Bracket Challenge. In our recent interviews with SAP senior research scientist Eric Blabac and SAP senior experience designer Anthony Jakubiak provided technical insights to the Match-Up Analysis tool, while SAP Brand Ambassador, NHL Hall of Famer, and NBC hockey analyst Eddie Olczyk gave his perspective on this statistical expansion.
Powered by SAP, the Match-up Analysis is an advanced predictive analytic tool featured in the NHL’s Bracket Challenge. Through a predictive model that utilizes a 37-algorithm system, a prediction is made for each series. With the intention of increasing fan engagement, SAP and the NHL’s tool gives fans a quantitative resource to help inspire their brackets.
Get The Latest Sports Tech News In Your Inbox!
To get to this final build of their tool, SAP and the NHL experimented with numerous iterations. Blabac described the process, which required collecting essentially all of the potential variables that contribute to a playoff victory. Between SAP and the NHL, over 240 variables were initially run through statistical models. Eventually, it was narrowed down to 37 variables that collaboratively had the best predictive power. Two kinds of factors were considered, regular season factors (including penalty kill percentage, points percentage, save percentage, etc.), and in-series factors (such as home or away location, how many time zones the team is traveling, and performance in last five and fifteen games).
Jakubiak provided insights into the design iterations to reach the final product. The approach to this project was similar to Phase 1 of SAP’s statistics overhaul. There were over thirty iterations prior to reaching the final design. SAP worked very closely with fans and sketched out those experiences to create the final design.
According to Jakubiak, the design provides enough context to the predictive results, rather than simply showing which team has the overall edge. SAP wanted to allows fans to visually see the statistics that supported each prediction decision by showing which team had the advantage per category, further engaging fans with the system and statistical approach.
While this playoff predictor was only intended for fans, it certainly could prove useful for teams by allowing them to get an understanding on what particular factors they would need to win playoff games. The current system is currently designed at a team level, rather than a player level. At some point, a similar system could be utilized for fantasy sports.
Fans may compare this predictive system to the already established predictor EA Sports provides each season. While SAP hopes to have as accurate of a system, the processes are completely different. SAP’s system has a statistical base approach, considering past performances and predictors. On the other hand, EA has more of a simulation approach at the player level. EA’s simulation is performed prior to the start of a season. Conversely, SAP’s predicts once the regular season ends.
Internal testing against the previous four seasons attributed to its preparation. The 2014 Stanley Cup Playoffs were tested to observe the accuracy of the prediction system. The Los Angeles Kings were accurately predicted as the 2014 Stanley Cup Champions, although their opponent in the Stanley Cup Final was predicted to be the Boston Bruins, rather than the New York Rangers.
At the game level, the Match-Up Analysis is showing roughly 80% accuracy. As exemplified by this tool, no predictive system can be 100% accurate, which begs the question— will there be any issues with credibility in the 20% chance of inaccuracy? Blabac said “Every fan knows anything can happen in the Stanley Cup Playoffs.” Meaning every fan is aware of the unpredictability of these games. The intention of this tool is to assist fans in understanding which team has the edge in a series.
From a credibility standpoint, Blabac stated how fans will see the accuracy of the system moving forward into the playoffs. Jakubiak added onto Blabac’s thoughts saying how the purpose of this tool is to “Engage fans with the context of the stats,” especially fans that are not “hardcore stats fans.” This statistical approach to the playoffs allows fans to dip “their toes into the vast amount of stats that the NHL has and the sport has around the game,” rather than necessarily perfectly pick a winner to each series.
In the case of the Match-Up Analysis’ selected team being eliminated, there will still be a system in place allowing fans to continuously see which team will have an edge for the next series. On the NHL’s website, predictions will be rerun every night based on that day’s finishes and update along the way. For the Bracket Challenge, every single combination possible will be statistically analyzed, so regardless of who is eliminated, every match-up will be predicted by SAP’s tool.
Blabac and Jakubiak both expressed excitement over the latest development of the NHL’s and SAP partnership. Eric Blabac said he was excited about “the whole thing, the content is very exciting and new,” explaining how this statistical approach to the Stanley Cup Playoffs “gives fans a better understanding in the playoffs. It’s a huge step forward and great to be a part of it.” Jakubiak echoed Blabac’s sentiments saying how this build off of Phase 1 is “Another tool set for fans to engage and dive into the game.” The deeper engagement allotted to fans by Match-Up Analysis allows “fans to see the wonderful sport that hockey is.”
NHL Hall of Famer, NBC hockey analyst, and the latest SAP Brand Ambassador, Eddie Olczyk gave his hockey-minded perspective on Phase 2 of the NHL and SAP’s partnership. Olczyk, a self-proclaimed “traditional guy” when it comes to analytics in hockey, described the playoffs as “the best time of year.”
Olczyk explained how statistics that go into every matchup help quantify the strengths and weaknesses of a team. However, there are teams that defy their typical numbers and simply play well against particular opponents. Those unquantifiable factors must be considered, regardless of what the numbers say.
Advanced scouting and analytics may be a part of the game now, but Olczyk still insists upon the “eye test” of how teams are playing. “I know what the numbers say and what key statistics say, you think how teams rely on goaltenders would catch up with them.” Olczyk explained, “Numbers are what they are, a very good indication on how a series may go, but not necessarily who wins a particular game. Teams are who they are, numbers say that over a regular season… As a broadcaster, I have the information, the teams and coaches have information, but the execution or lack thereof in a particular game or playoff series— that’s the deciding factor.”
Having said that, Olczyk recognized the potential of a tool like the Match-Up Analysis. As a former player and coach, this absolutely is a tool Olczyk would have utilized. He believes the information available to coaches and organizations is incredibly advanced now. But is implementing more technology into hockey negative for the sport? Olczyk says that ‘negative’ may not be the right word, since “as a coach, you can’t have enough information” and advanced statistics allow coaches to have information at their disposal. All of the enhanced statistics can help a coach by indicating how to create the most collectively advantageous situation for a team. But, according to Olczyk the challenge for coaches, players, and fans is deciphering exactly which information is useful.
Some teams though, defy advanced statistics and continue to have success while their numbers say otherwise. The Montreal Canadiens, Minnesota Wild, and New York Rangers are all teams that are not properly represented by their analytical numbers. The Los Angeles Kings are another team with intangibles that allow them to excel in the playoff atmosphere, despite their (often) lackluster regular season numbers. So fans may question whether the predictor will properly represent those teams. Olczyk said that “goaltending camouflages a lot of deficiencies… the goaltender is the greatest equalizer in sports.” So while it may appear that teams like the Canadiens and Rangers have shortcomings elsewhere and often have to rely on their goaltending, Eddie Olczyk says “there is greater risk in your end, but I would take my chances with the Canadiens and Rangers.”
Other reminders for those filling out their brackets— “Teams play different come the playoffs, they do, regardless of what anybody says.. the intensity goes much higher.” Olczyk also stressed how adaptability is a crucial factor for teams. “I think that’s another thing, you get to this time of the year and the game is different… That’s why some of the analytics do not necessarily bear out… And having won a Stanley Cup, you need some luck too somewhere along the line.”
The partnership between the NHL and SAP in Eddie Olczyk’s opinion have “taken the game to today” with the influx of instantaneous information. So for a more traditional-minded NHLer, Olczyk still relies on his instincts, but considers the information provided by the numbers that break down all of the strengths and weaknesses of teams. “I think it brings the NHL current when it comes to numbers and analytics, much more relevant for teams and fans alike because its a new way to look at the game.. It takes people inside the game. You can go as inside as you want and you can break it down however you want and come out with a winner in a series.” And from a broadcaster’s perspective, Olczyk sees the dividends of having instant statistics to discuss during a game.
Overall, Olczyk looks at how the NHL is phasing hockey into today, “We’re very early in the phases moving forward with the Match-Up Analysis, but I think it’s the way the world is now and the NHL has caught up and been proactive.” The NHL’s Stanley Cup Playoffs Bracket Challenge begins today and simultaneously the Match-Up Analysis tool will become available for fans.