Alexander Ovechkin, lies, damned lies, and statistics

Jonathan Willis
May 13 2012 12:04PM

Photo via Wikimedia Commons, GNU Free Documentation license

Photo via Wikimedia Commons, GNU Free Documentation license

There was a relationship that came up a few times during Washington’s post-season run, ended in (yet another) one-goal game last night against New York. When Alexander Ovechkin played more than 20:00 in a game, the Capitals were 1-and-5; when Ovechkin played less than 20:00 per game, the Capitals were 6-and-2.

Is that useful information?

No, it isn’t. Rather, it’s a case of lies, damned lies, and statistics.

That quote, popularized by Mark Twain, refers to the way that a statistic, while being factually accurate, can mislead.

Taking the numbers from the opening paragraph, and no other information, we might conclude that Washington has a better chance of winning with Ovechkin playing reduced minutes. Certainly, there’s a strong correlation between Washington winning and Ovechkin playing less. In this case, however, correlation does not equal causation.

Here’s another strong relationship: the one between scoring first and winning a hockey game. In the 2012 playoffs, Washington went 7-and-1 when scoring first. They lost all six games they played when the opposition scored first. Scoring first, as has been shown time and again in the past, has a strong relationship to winning. This is especially true for teams like Washington – where the coaching strategy is essentially to score one goal and then bunker down for the rest of the game.

Washington doesn’t need Ovechkin on the ice to bunker down. His talents lie more to the offensive side of the puck.

Unsurprisingly, then, the games where Ovechkin played the most minutes also tended to be the games where Washington was trying to come back from an early deficit. Of the six games where Ovechkin played more than 20 minutes, four of them were contests where Washington was scored on first. Of the two remaining, one was a lengthy overtime game, while the other saw Washington hold the lead for just a dozen minutes of the game. Ovechkin played when the Capitals were behind on the scoreboard; not surprisingly, the Capitals tended to lose when they were behind on the scoreboard.

Of the eight games where Ovechkin played less than 20:00, six of them were contests where Washington scored first and spent most of the night defending the lead.

Without knowledge of Dale Hunter’s strategy, knowing that Ovechkin plays more in losses is highly misleading.

74b7cedc5d8bfbe88cf071309e98d2c3
Jonathan Willis is a freelance writer. He currently works for Oilers Nation, Sportsnet, the Edmonton Journal and Bleacher Report. He's co-written three books and worked for myriad websites, including Grantland, ESPN, The Score, and Hockey Prospectus. He was previously the founder and managing editor of Copper & Blue.
Avatar
#1 Derek Jedamski
May 13 2012, 12:16PM
Trash it!
0
trashes
Props
0
props

This relationship between correlation and causation is one that is not easily understood by too many people. It's laid out here in a nice clear manner, good read.

Avatar
#2 W. Ron Sweeney
May 13 2012, 12:39PM
Trash it!
0
trashes
Props
0
props

So good. This "statistic" was figuratively killing me. Can we require a statistics course for all NHL commentators? And introduce some of them to score effects. My pet peeve is the "when player X scores, his team wins." Is there anyone whose team has a losing record when they score?

Avatar
#3 SmellOfVictory
May 13 2012, 02:35PM
Trash it!
0
trashes
Props
0
props
W. Ron Sweeney wrote:

So good. This "statistic" was figuratively killing me. Can we require a statistics course for all NHL commentators? And introduce some of them to score effects. My pet peeve is the "when player X scores, his team wins." Is there anyone whose team has a losing record when they score?

I'm sure it's possible, if you have a prolific enough scorer on a bad enough team (maybe Kovalchuk on the Thrashers?)

Avatar
#4 SmellOfVictory
May 13 2012, 02:37PM
Trash it!
0
trashes
Props
0
props

Ooh, and Robyn Regehr definitely had that, at least one year.

Avatar
#5 breaklance
May 14 2012, 09:18AM
Trash it!
0
trashes
Props
0
props

Only thing you ever need to know about statistics - they can say whatever you want them to.

Say your testing medecine, the first batch had a mortality rate of 68% but you won't say that to the board of directors who's giving you money. You'll say it had a 90% success rate in the first 2 weeks of testing(which for the sake of this example is true). Even though after that people started dropping like flies, your statement is still technically correct.

Avatar
#6 Tach
May 14 2012, 10:27AM
Trash it!
0
trashes
Props
0
props

The easiest way to explain this phenomenon to people in my experience is QB kneel downs in football. The team who has the QB take a knee the most wins something like 98% of the time, but even the thickest clod can pick up that the QB taking a knee is because they were winning and not vice-versa and just having your QB take a knee over and over again is a sure route to defeat.

Comments are closed for this article.