On Tuesday, Capitals beat writer Isabelle Khurshudyan held a wide ranging interview with General Manager Brian MacLellan and Head Coach Barry Trotz discussing how they’ll improve the team, be it through “statistics, video or, as Trotz put it, ‘the eyeball test.” Since the season is almost here, it’s time to dip our feet into the analytics versus eye test debate.
The Eye Test
This is exactly what it sounds like. Players and coaches watch game film to determine the systems of other teams, as well as strengths and weaknesses within their own. From there, coaches help the team implement the right system to use against their opponent and help players with individualized attention to problem areas in their game. Craig Laughlin does this when he works his telestrator magic. Let’s do an eye test of our own:
- Marcus Johansson steals the puck from Jakub Voracek, and Voracek crashes and burns. Johansson then crosses the Flyers blue line bringing a defender with him.
- Andre Burakovsky drives hard to the front of the net, taking defender Radko Gudas with him. This creates an opening for Jay Beagle, who is able to get to that space unchecked because:
- Shayne Gostisbehere is going for a change.
- Claude Giroux is gliding back to aid the defense instead of recognizing this as an odd man situation in favor of the Capitals.
- Poor Brandon Manning (#23) positions his stick in the passing lane where Johansson would pass the puck to net-front Burakovsky. Giroux should be screaming at him that Burakovsky is covered by a back checking Gudas, but Beagle who is at his one o’clock is wide open, and should therefore move his stick.
- Beagle receives a pass from Johansson in the high slot and rips it top shelf. Gudas tries to block the shot and fails.
Both problems are solved with a few head swivels and situational awareness, and that goal never happens again.
But the problem with the eye test is that it is just that; your eyes, if you’re a player or coach, you can only see so much. Brandon Manning in that previous goal is playing textbook defense: his skates are pointed towards the puck possessing player, his stick is in what is typically the most dangerous passing lane for that scenario, and his gap control is perfect. Manning just can’t see the player behind him.
Additionally, most players look good when they play but make subtle mistakes that cost teams games or drag other players down. They can be kept in lineups either due to coaching bias (Tanner Glass of the NY Rangers) or because there is an older stat that can be inflated to make that player look good. Remember the year Jeff Schultz led the Capitals in plus/minus, even though the main reason for this was because mostly he was on the ice with Alex Ovechkin, who had the 2nd highest plus/minus on the team? Your eyes deceive you sometimes.
The concept of advanced statistics/sabermetrics/analytics is originally from baseball. The Hollywood blockbuster Moneyball, based on the Michael Lewis book, popularized this concept by focusing on the cash strapped Oakland Athletics, who had to find a way to replace the talent they lost to big money clubs by focusing on the statistics that correlated to wins. The basis of their theory was that if a player gets on base in any fashion, that leads to more runs, which leads to more wins. They gamed the system and made it to the playoffs that year, and it cost them less for a win than the New York Yankees or Boston Red Sox.
When Wayne Gretzky said “you miss 100% of the shots you don’t take”, he inadvertently gave us hockey’s first analytic tool: shot attempts. Here’s the thinking behind shot attempts as a primary source for analytics: if your team is shooting more on your opponents net, you have the puck more. Therefore, you should score more since the other team doesn’t have the puck. Circular thinking for sure, but crafty.
When we talk about analytics, these are typically the stats being discussed. These statistics do more heavy lifting in terms of finding players who are underperforming or playing very well on a very bad team. For example, we know that last season’s pick up of Justin Williams was a smart move because he elevated the play (puck possession numbers) of everyone around him. And finding those kinds of players, especially ones who are younger and on entry level contracts, can get you cheaper wins than trying to trade for Anze Kopitar. If you’re wondering why the Coyotes hired an analytics whiz kid as their GM, that’s a pretty good reason.
But it’s important to proceed with caution. If you’re using analytics, the following apply:
- Correlation does not equal causation. While stats like Corsi and Fenwick can measure a player or teams possession numbers, the context for those numbers matter, and just because a player or team has good possession numbers doesn’t mean they’re talented.
- Hockey is a team sport which means every interaction with a teammate or an opponent is important. Until we have individual player tracking during every single game, getting to the absolute root of these statistics is difficult.
Analytics does have its flaws. There aren’t great numbers yet specifically for defensive defensemen or goalies, and while those are being developed, we still don’t really know what good defense looks like analytically. For example, we know that Erik Karlsson puts up gaudy offensive numbers, and that he’s a Norris trophy candidate. But what makes him a good defenseman in his own zone? The eye test is really good for him because he completes most of his first passes out of the zone, or carries the puck well through the neutral zone.
In summary: more info means better decision making, and you should be using both the eye test and analytics when judging players or teams. Good thing the Caps are using both.
By Julia Karron