SAP Develops NHL Playoff Bracket Competition Just In Time For Playoffs

Following the development of the NHL.com website’s new Stats homepage, SAP has again collaborated with the NHL in developing the Bracket Challenge, which will encourage fans to make their playoff picks with the help of advanced analytics, just in time for the playoffs.

As opposed to impromptu bracket picks or informal office pool predictions, SAP’s Match-up Analysis Insights will allow fans access to the kind of advanced statistical metrics that tend to pick winners, as opposed to random guessing, or gut feelings, or declaring hunches based on whimsical factors like jersey colour or mullet size.

Powered by SAP’s HANA Enterprise Cloud service, the Bracket Challenge will examine each series based on a predictive model that incorporates 37 variables, including road record, goals against per game, penalty kill percentage, points percentage, and shot attempts percentage behind.

Speaking with SAP Senior Research Scientist Eric Blabac, Cantech Letter asked how the Bracket Challenge expands on SAP’s already existing Stats collaboration with the NHL and what technical factors go into shaping an analytical approach to predicting hockey outcomes.

“What we’ve actually developed is statistical models based on historical playoff data and really understanding which factors help a team win a playoff game,” said Blabac. “We went back four or five years, to the event-level data, and found 37 factors that play into which team is going to win a playoff game. Using these 37 factors, we then take all the games, predict each game, and then predict the rest of the bracket, as well. These are legitimately quantified odds that we’re putting on the site.”

Fans posting about their bracket picks on social media are encouraged to use the #NHLBracketChallenge hashtag.

“SAP is providing fans with extensive analytical resources and new stats at NHL.com to help fans fill out their brackets while Samsung is offering innovative, new to market products for the winning entries,” said NHL Vice President Product Development and Social Media, John Pacino. “Without question, the Bracket Challenge has proven to be an effective way to increase engagement among avid fans and to connect with a wider audience of hockey fans.”

In addition to competing against friends and colleagues, fans can test their predictive mettle against the “Famous Hockey Fans” group which includes actors Michael J. Fox, Jon Hamm and Taylor Kitsch, as well as actress Alyssa Milano. Last year, Milano was crowned champion of the “Famous Fans League,” correctly guessing that the Los Angeles Kings would win the Cup, a feat accomplished by fewer than 1% of all entries worldwide. Alyssa Milano, hockey savant. Who knew?

As of now, fans can register at NHL.com/Brackets to predict outcomes and challenge their friends for prediction supremacy bragging rights. On April 15, the brackets will be set and the actual playoffs will begin, at which point fans can check their own prediction accuracy, or lack thereof, against the actual outcome of games played.

FAN DEBATE: The stat that shows which team may win the #StanleyCup Final. Watch and discuss.https://t.co/aP9XYpyWNf
— NHL (@NHL) April 8, 2015

About The Author / Terry Dawes

Comment

Darrell Grob says:

April 14, 2015 at 6:18 am

First, it’s convenient that they could cherry pick that the last whatever cup winners were in the top four. Second, the very fact that only three of the last seven cup winners were ranked first in Corsi rating shows that it is not a reliable statistic. It’s a telling statistic, but not reliable. And their disclaimer at the end of the clip of “anything can happen” is certainly convenient. Of course, anything can happen!
I read another article where SAP was touting that they accurately predicted the Kings to win the SC. Well, duh. A lot of people without computerized stats predicted that. They also said they missed the eastern conference contender. They predicted Boston. So they were only 50% right on who would be in the finals. Not a blazing hot result.
I actually like statistics. I’m into them. But I’ve learned that season long statistics are only worth about 50% of their own weight. The reason? Because for a top rated team, you’re playing a lot of patsies. (BUFF or ARIZ) Statistics get padded against teams like that.
Now, I would buy into this statistical model more if it were only for the last 10 or 20 games.
Another interesting stat from last year. (Disclosure. I’m a big Blues fan.) Last year they smote the eastern conference which probably helped their stats. But they couldn’t get past the Blackhawks, again. Interconference statistics are meaningless in conference playoffs.
Then there’s matchups. In the upcoming matchup of the Blues and the Wild (a very exciting team!), what if Tarasenko lights up Dubnyk like a Christmas tree because Vlady knows were Dubnyk is soft? Or what if the Blue’s defense can stand the Wild offense up at their own blue line? And just the opposite could be true for Minnesota. What if Parisi take’s Allen’s pants off on the ice. Or Dubnyk is Superman with pads on?
Like I said, I like statistics, but for the reasons I’ve stated, I think this stat, at least how it’s presented, is soft.
Thanks for reading.