Hockey Analytics and What They Mean for the CapsPosted on March 27, 2014 by Ben Bruno
Yesterday at the University of Alberta, a presentation titled "Hockey Analytics: The new wave of information and the online fan community that is driving the field" was given by Sunil Agnihotri and Michael Parkatti. The purpose of their presentation was to give a basic overview of "advanced stats" in relation to hockey and how they can be used for things like determining future outcomes, as well as talent evaluation. I highly recommend viewing the material and I have added a recording of the presentation at the end of this post. Parkatti provided his analysis in an easy to understand manner and let's see how it relates to the Capitals.
Parkatti introduced the idea of advanced stats as being data that "strips out the inherent luck of goal" scoring. So many things need to happen correctly for a player to score a goal, and many of them are based purely on "luck", such as a puck bouncing off of a defender and into the net. Advanced stats are used to eliminate this luck and focus more on tangible information that can then be used to find patterns or determine outcomes.
One of the first things that Parkatti talked about was how a team's points percentage and goals scored percentage are useless when trying to determine current versus future performance. He went on to say, however, that Corsi% can be used effectively to determine this and that at the team level, Corsi% and Fenwick are "much better predictors of future performance than things like goal percentage or even how many points you have right now in the standings." Check out his chart below for a visual of what he's saying.
Back in November of last year, I wrote two posts on the importance of play at even-strength and how it is indicative in telling which teams will and will not make the playoffs. On November 21st, the Caps' Fenwick for percentage (FF%) was 47.9. Today, four months later, their FF% is 48.3%, barely improved. I revisited the topic of their 5v5 play three weeks ago in another post and came away with the conclusion that the Caps "are who we thought they were", since their play at even-strength had stayed relatively the same all season. Based on Parkatti's chart above, it looks like I was unfortunately right.
Parkatti also spoke about Corsi at the player level and how it "is going to be a much better predictor of your future goal% than your current goal%" and is the reason why I selected a picture of Dustin Penner to accompany this post. Penner has the highest CF% of any forward for the Capitals, yet he is being used sparingly on the 4th line. Granted his time on the ice for the Caps is a fraction of what most of his teammates are, but he has still performed well in his role while playing with less talented players. I can only imagine what his numbers would look like when playing in the top-6, as opposed to skating with the likes of Tom Wilson and his 46.5 CF%. Instead, Oates continues to use Jay Beagle on the first line, but that's another story.
The next topic Parkatti discussed was the relationship between Corsi and teams that won the Stanley Cup (see chart below). He mentioned that there's "only three teams that have won the Cup in the last 15 years, that in the regular season had a shot share that was less than 53%." This should be a frightening statistic for Capitals fans, as the Caps are 5% off of this number, which may not seem like a lot, but over the course of a season is a huge gap. The three teams that are the exception are Pittsburgh, Boston, and Carolina. Boston had Tim Thomas in net the year they won it all and he had a historic playoff run, while both Pittsburgh and Carolina had "incredibly high shooting percentages" in the years they won, according to Parkatti. Unfortunately for the Caps, their team save percentage is good for 19th best in the league, while their shooting percentage sits at 20th best, meaning that they won't be able to rely on either to help make up for their subpar Corsi%.
The next topic discussed was shot quality and if it exists or not. While the data aren't completely conclusive, Parkatti mentions that there is some truth to the idea of crashing the net. Shots from 26 feet and closer go in at a rate above the league average, while shots from 27 feet and further go in below league average (see chart below). Taking this a step further, Parkatti showed that while just 6.3% of all shot attempts end up as rebounds (meaning another shot attempt was created within 3 seconds of the original shot attempt), they account for 23.3% of all goals scored. In relating this back to the Caps, this again points to giving Penner more playing time. He is a big-body that plays close to the net and could surely benefit from some rebounds off of blasts from Ovechkin.
The last topic I want to mention relates to goalies, and specifically Braden Holtby. Parkatti mentioned that it takes roughly 3,300 shots against to get a true sense of the talent level of a goalie in the NHL. This wasn't a major point of anything Parkatti was discussing and rather he was using it as a qualifier to another point, but it stood out to me since Holtby has only faced 2,936 shots against for his career in the regular season. While he isn't far off of Parkatti's number of 3,300, it should be noted that the book could potentially still be open on Holtby, and unfortunately for him, with Halak around, the book will remain open until we see who the starter is next season. Not really an "advanced stat", but noteworthy.
Parkatti also mentioned how much of the advanced stats world in hockey is user driven and has started to gain some mainstream momentum because of the internet world and social media, so while this information might be new to you, it will not be long (hopefully) before it has a firm place in hockey. Some of the statistics are proven to be effective means of both predicting the future and evaluating talent, and while statistics are by no means the be-all-end-all, they deserve their place in hockey, right alongside with the traditional scouts and "eye-test" evaluators. A combination of the two is necessary in today's game if team's want to put themselves in the best possible position to succeed. Fancy stats are the wave of the future so jump on the train now and check out how the Caps match up to other teams in the NHL, even if their numbers make you cringe.
Agnihotri and Parkatti's Presentation: