My Anti-Advanced Analytics Rant: Trust yourself, you know the numbers

ST. LOUIS, MO - FEBRUARY 16: Pheonix Copley #30 of the St. Louis Blues talks to goaltending coach Jim Corsi prior to playing against the Dallas Stars at the Scottrade Center on February 16, 2016 in St. Louis, Missouri. (Photo by Scott Rovak/NHLI via Getty Images)
ST. LOUIS, MO - FEBRUARY 16: Pheonix Copley #30 of the St. Louis Blues talks to goaltending coach Jim Corsi prior to playing against the Dallas Stars at the Scottrade Center on February 16, 2016 in St. Louis, Missouri. (Photo by Scott Rovak/NHLI via Getty Images)

Advanced Analytics is all the rage. I’m not a hater of technology or math, but common sense and logic are being overrun by numbers with shaky foundations.

Generally, I’m not a ranter, but I’ve got something to get off my chest. I have a problem with advanced hockey stats. My issue isn’t that they’re too complicated, or I don’t like math. Just the opposite, although I think the primary metrics are a little convoluted, I understand the math.

To clarify I don’t dislike all advanced stats, just some of the primary ones. I’m going to give you my reasons in just a minute. Keep in mind I know it’s 2018 and I know everything is quantified and cataloged.

More from Pucks of a Feather

Possession is Overrated

The two primary advanced stats are Corsi and Fenwick, which both measure puck possession. The prevailing wisdom of those metrics is the more a team possesses the puck the better chance they have of winning. They are measured by a ratio of shot attempts to shots against the difference is Corsi counts shots and missed shots while Fenwick counts shots, missed shots, and blocked shots.

My problem with Corsi and Fenwick is that I don’t believe the correlation between puck possession and winning is as strong. I’ve watched too many games decided on turnovers, odd plays or even one good or bad shift.

Great players like Wayne Gretzky, Mario Lemieux, and Connor McDavid have changed games by picking a pocket skating down the ice at amazing speeds and putting the puck past the opposing goalie. Anaheim Ducks fans have seen Andrew Cogliano do it several times over his career.

My point is that too many factors happen in an NHL game to quantify it by puck possession.

Puck Luck?

PDO is an interesting metric, but one that I find seriously flawed. It supposedly measures how lucky a team is. A teams shooting percentage and save percentage are combined together and it assumes all teams should average 1.000. If the PDO is higher, a team isn’t as good as they appear and vice versa.

Well, sometimes a team is that good or that bad. Also, every team has its hot and cold streaks. This metric more than others aren’t an accurate reflection of how a team is playing over a short period of time.

I remember a statistics class I took in college. The first day the professor asked us if she flipped a quarter 100 times and it came up heads, how many people would bet on tails for flip 101? I raised my hand and in an auditorium full of students, she asked me why I raised my hand. I said that it’s a 50/50 shot and tails was due. Her answer changed my point of view when she said, “the quarter doesn’t know tails is due.”

Eye Test

This is So Cal so I’m busting out a Vin Scully quote.

“Statistics are used much like a drunk uses a lamppost: for support rather than illumination”

If you watch enough of a player or team, any statistic shouldn’t be a surprise. For example, if you watched the Anaheim Ducks over the last few seasons you know that Ryan Kesler was still a better faceoff man than his opponent more often then not. You also know that last year he wasn’t as dominant as in years past. You saw it with your own eyes and the specific percentages should be about what you think they are.

Do you need a metric to tell you that the Ducks take more defensive zone faceoffs than their opponents? Nope. You’ve seen one of the most penalized teams suffering the consequences.

Paying attention will still be your best way to judge.

How they’re used

There you are talking to some of your buddies and one them says Hampus Lindholm had a “Corsi For” of 51.8. That’s interesting if you know what Corsi is, but the problem is by itself the number means nothing. The only context it has is in comparison to another player. Even then, it isn’t informative unless it’s used in regard to someone who plays the same position.

Just throwing around a Corsi or Fenwick number around has no weight. Yet fans do it all the time. There are some advanced stats I do like. Win Shares and Zone Starts are two, but they don’t get as much attention.

I appreciate the chance to rant.