Hockey Math Works, Part II

I was taken to task for calling the logic behind the predictions in this article "math", and there is merit to the argument. Essentially, what I did was use logic and reason in making observations about statistics in order to make predictions based on the concept of reversion, or regression. I certainly didn't do any brute force math, that was all figured out by fellows like the man on the right, Sir Francis Galton.

What I did, however, was understand mathematical concepts and use basic logic and reasoning to make predictions. It wasn't groundbreaking, but at the very least, those predictions demonstrated the power of using those basic concepts and logic to further a fan's understanding of the game.

Way back in October, Oilers Nation held a contest in which they asked the contributing writers to predict the Oilers top 13 scorers and predict their season-ending goals, assists and total points.

Robin Brownlee went with 11 Oilers over the 40 point barrier, including 9 forwards. Jonathan Willis went with 5 Oilers over the 40 point barrier, all forwards.

Tyler Dellow took Brownlee to task for this because it defied common sense:

Something has to give though, in terms of ice time - someone has to take the faceoffs in the Oilers’ end, someone won’t get the necessary PP time (Brownlee has all of Horcoff, Hemsky and Penner at or near career highs in points, which implies that they’re getting lots of PP minutes, I don’t understand how guys like Cogliano are going to score 40+ in those circumstances). I made the (roundly ignored by those who dumped on my argument) point that there’s an opportunity cost to players like Cogliano being shunted aside so that we can worship at the feet of the Holy Trinity; Brownlee apparently doesn’t think that there will be.

Mr. Dellow was using simple logic - even if the Oilers stayed healthy, there just weren't enough minutes to go around. Note that there is nothing to do with advanced math in that line of thinking.

Put it another way - if the Oilers had lived up to the prediction, it would have made them the deepest offensive team in the history of the NHL, using 40 points as the cutoff for scoring depth. The 1980 Blues had 10 forwards score 40 points and the 1970 Bruins had 9. The 1974 Canadiens had 8. Adjust for era effects and Mr. Brownlee predicted that a team coming off of a 30th-place finish would have one of the greatest team offenses in NHL history.

Mr. Brownlee also predicted Gilbert Brule would break his career high in points by 30%, despite warning signs all over the place. On the other hand Mr. Willis predicted a 14% decline in Brule's point totals based on his own warnings about the upcoming season.

In the end, the Oilers had 4 players score 40 points or more. Gilbert Brule reverted to his former self.

This wasn't written with the intent of calling Mr. Brownlee out. Rather, I hope this was instructive. Yes, the collective media now has unparalleled access to players and coaches. Yes, they have access to the locker room, practices and upper management. No, that doesn't mean that the media always uses that access to understand the basic math and logic behind the game. At times, they rely on hubris and intuition, as Mr. Brownlee openly admitted:

With all the advanced statistics websites out there the disciples of digits who lean heavily on them to predict future performance should clean up on rubes like me when it comes to a simple hockey pool, no?

It stands to reason that if a person can run the numbers, extrapolating this and projecting that based on a bunch of formulas I don't understand, or really care to, surely they've got a leg up on an old hockey writer who simply eyeballs the players, glances at what nerds call the boxcars and throws out his predictions.

At the very least, the advanced stats people are making a more educated guess than the saw-him-good guys (more nerdspeak) like me, no? All this QualComp and Corsi and Desjardins jazz has to be a big advantage up against somebody who thinks taking into account even-strength ice time and power-play minutes is getting in really deep, no?

I don't think so. Not for a minute. No sale.

Mr. Brownlee used his eyeballs and was off by country mile. Mr. Willis filled the gaps with math and logic and came darned close. I'd call that cleaning up on the rube.

It's worth remembering that the depth and breadth of one's hockey knowledge has nothing to do with "proximity to the players".