Tyler Bozak’s having a productive few games

Let’s talk about Tyler Bozak for one second:

The most points Tyler Bozak has ever earned in one nine-game stretch of his National Hockey League career is 12, and that was the nine-game stretch ending with that game against Boston Tuesday. While he was held off the scoresheet Wednesday, he still scored the shootout-winning goal. Since returning to the lineup, Bozak has been held off the scoresheet just twice, has two three-point nights, and has two shootout winners to his credit. More importantly, the Leafs are 5-4-0 since his return, obviously spurred by his attendance.

Question: has Tyler Bozak turned the corner? Answer: Probably not.

I don’t mean to dampen anybody’s mood. After all, this is a happy day at TLN. The sun is shining into the blog’s Vancouver headquarters (oh dear, is it still snowy in Toronto?), the team has won three consecutive games and we’ve just introduced a couple of new writers to our team. The only thing left is for “that Cam idiot” to come here and crap all over Tyler Bozak’s hot streak by suggesting it’s a small sample, or unsustainable, or whatever it is.

And that’s exactly what I’ll do. 

(Before anybody stops me with a “but the team has played sooooo much better since Bozak came back to the lineup” I’m just going to remind you that the Leafs lost six straight games with Bozak in the lineup before the games against Carolina and then the Winter Classic. Toronto are 12-11-2 with Bozak in the lineup [an 85-point pace] and 12-9-3 when he isn’t dressed [a 92-point pace]. While it’s a flimsy argument that Bozak makes Toronto worse, it disproves that Toronto is “better” when he’s around if you’re looking at wins and losses totals to increase your argument.) 

Just to make the point, we’re going to bring in an old friend from last year, Nazem Kadri’s on-ice shooting percentage:

On-ice shooting percentage isn’t particularly advanced. It’s on-ice goals divided by on-ice shots, taking into account only 5-on-5 play. It’s also highly unpredictable and unstable, and being a point-a-game player thanks to having an on-ice shooting percentage does not mean a player is a point-a-game player. 

I think a lot of observers that tend to worry about the numbers and not the game may note that Nazem Kadri has had a bad year, or at the very least, a year where he has failed to improve. Not really. Taking into account defence, setups, overall offensive involvement and two-way play I’d suggest Kadri has improved, even if his points-per-game rate has dipped by 33%. That was to be expected, though. I hammered it home on this website all summer long.

Bozak has 11 points in nine games. That’s happened one other time (if we discount all the games from this little run of his) between November 10 and November 27, 2011. Over his next nine games, from November 30 to December 19, Bozak had just seven points, and over the nine games after, he had just three.

It’s difficult to ignore the volume of words from the Bozak Brigade, particularly given everything I said about him in the offseason. The problem with this is that if you expect me to mention Bozak is playing great every time he goes on a hot streak (which every player does) then you should also expect me to rub it in when things go the wrong way. Over his career, Bozak has been most often observed scoring four points in a nine-game segment. Four points in nine games isn’t bad, but it’s not $4-million good. 

Adam Gretz at SB Nation got into all the math, if you’re into that:

For me, I don’t believe for a second that the math is all that necessary to know what’s going to happen in the long run. I bring up the analogy of chess with my friend Jeff, and even though he’s better than me, I still win 25-30 per cent of our games (although we haven’t played much since we now live in different cities). If we played ten games over the next ten days, and I happened to win five of them, knowing my history as often losing more than I win, would you be confident I would still win five games? I wouldn’t be. I’m very bearish on promoting anybody’s play based on a small sample of games (remember our Jonathan Bernier example from earlier this season), since I would get burned whenever I tried to in the past.

Regression to the mean is a fundamental concept for understanding how anything works. You may have heard the term “random walk” before, and the optimal picture would be a drunkard staggering down the sidewalk. If he gets too far off course, say, into the street, or bumping against the buildings he’s walking by, he’s going to deviate back towards the path, or the mean. Over the course of his mile-long jaunt from the pub to his front door, the “mean” or “average” will be a very smooth line, but the drunkard’s path looks to be more random than anything.

Because Bozak has been hurt for half the season, we can infer plausible conclusions from his play. The problem is that you can’t base your projection of a hockey player based on only seeing the good aspects. In 25 games, more is likely to go well right, or well wrong (just as our drunk may go an entire city block walking down the middle of the street. Eventually, he might correct himself). With 22 points in 25 games, Bozak has a points-per-game rate of 0.88 despite a career rate of 0.59. Compare the likelihood that Bozak suddenly got up two weeks ago and decided he was suddenly going to produce like an NHL All-Star, versus the likelihood that Bozak is in the middle of a particularly lucky streak. Of course, the first option isn’t impossible as much as it is implausible and if offered a bet that between now and the end of the season Bozak’s points-per-game rate will be 0.88 or higher, I would definitely bet against him.

None of this changes that Bozak’s hot streak has given the Leafs a bit of life after looking their worst on the season in a four-game losing streak, but this post is mostly a warning. Don’t latch on to good production when you only have a small sample to work with, since you will more often than not be heartbroken with what follows.

Let’s play a simple game here. Guess what Tyler Bozak’s points-per-game rate will be between now and the end of the year. At the end of the season I’ll revisit this post and analyze the data.

My own guess is 0.64.