I happen to play on a beer league hockey team here in the Edmonton area. It’s populated with friends I’ve had since before I could walk, and other local chaps we’ve met along the way. We’re a rag-tag group that crosses the gamut of possible age groups, from guys in their late 20s up to some in their 50s. The skill level ranges from guys who played major junior, junior A, college right down to those who never went past house league growing up. I’m in the latter group, and I’m a goalie.

We do quite well for what we are — we won the league once a couple of years ago. However, we have an odd tendency to deviate our performance wildly, alternately seeming unbeatable and getting into losing streaks. We’re also pretty infamous for getting into high-scoring pond hockey type of affairs. Hey, I’m used to it, I’ve been playing with these sick freaks my entire natural life.

One of my best friends has a little brother, who’s been playing with us ever since he could basically hold a hockey stick. He’s about 7 years younger than us, so his hockey maturation didn’t really concern many of us much. That is, until his teens, when he grew, developed hands, and suddenly became better than any of the sad lot of us. Since then he’s ping-ponged around the upper echelon of the hockey world, playing in the WHL, CIS, and through a couple of semi-pro men’s leagues in North America (ECHL, CHL).

Through the better part of the last decade, he’s been able to play with us on various occasions. Whenever he shows up to play, it’s like we have a secret weapon on our team, as if He-Man just began talking about the Power of Greyskull. The kid is simply just better than everyone on the ice, and it’s obvious within a few seconds of his first shift. He dangles on the rush, he has a wickedly accurate wrister, and generally makes defenders look foolish. He also makes everyone seem to play better, as if we’ve all been emboldened by his mastery. He’s the missing piece of our hockey team — the doorway into Narnia.

——————————————————————

Our Edmonton Oilers hope they have found their secret weapon as well – a kid poached out of college named Justin Schultz. And he came to the home squad in a not dissimilar way as my friend’s little brother — he just kind of showed up one day and started playing. I can’t really fault the Oilers’ management for any of this (hell, they must have *at least* opened the door and said ‘hi’ nicely). It’s been a long while since the Oilers had a legit sharpshooter on the backend, at least since Greg Hawgood (Pris Chronger notwithstanding).

There’s a lot of hype for this kid, and I’ve been wondering how warranted it all is. I don’t think there’s any concern he’s not a talented kid. I mean, he did just rip 1.41 points per game from the service line in the AHL over half a season worth of work. In this post, I’d like to get to the bottom of the mystery, and try to answer the question of what kind of weapon the Oilers actually have up their sleeves.

To begin, I needed to find a comparable population of defencemen who served time in the AHL before establishing themselves in the NHL. Restricting myself to post 1994-1995 and players who played their first NHL season before the age of 26, I set out two scenarios where a defenceman could get included in my sample population.

- He played at least 20 games in the AHL in one season, and then the
**next**year played his first NHL season with at least 40 games played - He played at least 15 games in the AHL in one season, and then the
**same**year played his first NHL season with at least 40 games played

I picked those dates arbitrarily, but I wanted to get as large a sample size as I could. It ended up that 186 defencemen matched either of those conditions. Next, I ran a series of multiple regression models where I attempted to use different variables to predict what their point per game (PPG) level ended up being in the NHL. The following is what I consider to be the best:

**Predicted NHL PPG = # of years since 1994 * 0.006743406 + Played in AHL during lockout? * 0.122645602 + AHL PPG * 0.316923127**

Let’s look at each of these three terms in isolation:

**— # of years since 1994 * 0.006743406 —**

The idea behind this term is that the quality of play in the AHL has increased over time relative to the NHL. I have a hypothesis that advances in coaching and equipment have generified pro hockey players to some degree over the last 20 years. That is, replacement level players coming up from the AHL are better relative to their NHL peers compared to how they were in the past. I set a baseline of 1994, or the beginning of the first lockout. Every year that passes since 1994 sees defensemen being able to retain 0.007 more points per game in their first substantial season in the NHL. This may not sound like much, but the implication is that a defenceman entering the NHL in 2012 will retain 0.12 more points per game than his peer from 1994. This variable has a t-stat of 5.5, and is statistically significant.

**— Played in AHL during lockout? * 0.122645602 —**

This is a dummy variable (meaning it can only be zero or one). My thinking here was that players who played in the AHL during a lockout season were more likely to have been in that league only because they were locked out from the NHL, and otherwise would have played in the NHL. It follows that they should be able to retain more of their AHL offense in the NHL, because they were, on average, better players than other AHL players entering the NHL in the past. With a t-stat of 4.9, we can say that there is a statistically significant relationship here. So, if a player played his last pre-NHL stint in the AHL during a lockout year, he will score 1 * 0.123 = 0.123 more points per game in his 1st NHL season than a player who played during a non-lockout year (0 * 0.123 = 0). To run a quick gut check, here’s a table of all the defencemen who played their AHL stint during a lockout, and how they were able to retain their offense between the AHL and NHL:

All but Gonchar (95-95 lockout) played during the 2004-2005 locked out NHL season. You can see that, on average, they retained 85% of their offense between the AHL and NHL, which is much higher than the 60% average of the total population of 186 players. In fact, 5 of these defencemen actually increased their PPG rate in the NHL compared to the AHL. It’s obvious that many of these players would have played in the NHL had there been a 2004-2005 season.

**— AHL PPG * 0.316923127 —**

The most powerful predictor of how a defencemen will score in the NHL is how he scored in the AHL. This calculation shows that a player will retain 0.317 or 31.7% of his AHL PPG scoring rate in the NHL. This variable has a t-stat of 11.1, and is statistically significant at a 99.9(followed by 21 more 9’s)% level of confidence. You can see how this plays out by looking at histograms comparing the 186 defencemen’s AHL point per game scoring to their NHL PPG scoring and how it declines between the two:

For instance, 23 of the 186 players scored between 0.6 and 0.7 points per game in their final pre-NHL AHL season, while none of them scored between 0.6 and 0.7 points per game when they finally got to the NHL. There are many players who seemed like scoring Gods in the AHL who barely kept their heads above water in the show (eg Andy Sutton, going from 0.71 PPG in the AHL to 0.05 PPG in the NHL).

Just as an aside, if you were to run the correlation using ONLY AHL points per game as a variable, it would suggest that AHL defencemen keep 0.470 or 47.0% of their offence (this could be referred to what Lowetide commonly calls NHLE (NHL Equivalency). However, the disparity between this 0.47 and the 0.32 seen in our final equation is a powerful indicator that those few locked-out Dmen have such an affect on the final outcome, further validating the 2nd variable talked about above.

**Predicting Mr. Schultz**

So now, to the fun part and predicting how Justin Schultz will do in the 2012-2013 season. Our equation has a correlation (R-square) of 0.833, meaning that variation in our 3 variables explain 83.3% of the variation in NHL points per game (It is also statistically significant with an F score of 303. Trust me, that’s very good). If you take the following equation:

**Predicted NHL PPG = # of years since 1994 * 0.006743406 + Played in AHL during lockout? * 0.122645602 + AHL PPG * 0.316923127**

and input Schultz’ numbers, you will get:

**Predicted NHL PPG = 18 * 0.006743406 + 1 * 0.122645602 + 1.41 * 0.316923127 = 0.69**

So, our equation predicts he will score 0.69 PPG in the upcoming season. Here’s a table to show how many points that translates into for Schultz and a few other potential rookie Dmen:

The equation predicts that Schultz will score 33 points in the upcoming 48-game NHL season. Having a look at some of the other defencemen on this list, it makes you wonder if this class will be as strong as the one seen in the prior lockout above. There are likely defencemen who played in the AHL this year who will become stars that we have barely even heard of. I wonder who they are…

Here’s a list of all the players included in my sample:

## 4 Comments

Do you have a job anymore?

Jobs? Where we’re going, we don’t need… jobs. —> actual movie quote

Wow – a stat-head nerd boy who plays hockey and can really write!

Nice post, I really like your site.

Great article, but your equation didn’t consider age, and this is one of the biggest factors when considering future production.