Behindthenet Blog

[Originally appeared in the Nov. 16, 2008 edition of 'The Committed Indian', the unofficial Chicago Black Hawks program...]

Can you imagine baseball fans across the country getting excited about a 16-year-old shortstop? Football fans getting worked up because a freshman running back is dominating the varsity somewhere in Texas? Well, it happens all the time in hockey. Ever since a 16-year-old Wayne Gretzky stepped on the ice in 1977 and dominated players three years his senior in the Ontario Hockey Association junior league, there's been tremendous buzz around kids who would be a decade away from their primes in many other sports. Hockey fans are always looking for the second-coming of Gretzky: Mario Lemieux in 1981, Eric Lindros in 1989, Sidney Crosby in 2003, and now John Tavares.

There was so much excitement around Tavares that Canada's junior hockey leagues changed their rules to allow him to be drafted at age 14 and play just three days after his 15th birthday. Fans of bad NHL teams were already hoping they'd be bad enough to get the first draft pick in 2009 so they could pick him. He did not disappoint, scoring 45 goals in 65 games. The list of players who've put up numbers like this at such a young age over the last 25 years is a short one:

Name            Season   Age League PPG   Adj
John Tavares    2005-06 15.28 OHL   1.18  1.01 
Jason Spezza    1998-99 15.55 OHL   1.06  0.83 
Rob Brown       1983-84 15.73 WHL   1.16  0.69 

The table lists each player's age as of January 1st of that season. The PPG column lists each player's points-per-game, while Adj shows that figure adjusted to a 6 goal-per-game offensive environment. This adjustment is particularly important - without it, we would overstate the significance of performances like Brown's, in a league where teams combined for more than 10 goals-per-game. At any rate, Tavares was younger and better than other prodigies, who themselves were pretty good: Spezza has three 30-goal seasons in the NHL before age 24, while Brown scored 49 at age 20.

Tavares didn't disappoint in his second season, scoring 72 goals in 67 games. His company at that level:

Name                  Season   Age League PPG   Adj
Sidney Crosby         2003-04 16.40 QMJHL 2.29  2.04 
John Tavares          2006-07 16.28 OHL   2.00  1.61 
Vincent Lecavalier    1996-97 16.70 QMJHL 1.59  1.27 
Derek Roy             1999-00 16.66 OHL   1.32  1.13 
Pierre-Marc Bouchard  2000-01 16.68 QMJHL 1.42  1.10 
Brian Bellows         1980-01 16.33 OHL   1.76  1.09 
Jimmy Carson          1984-85 16.45 QMJHL 1.71  1.05 
Martin Lapointe       1989-90 16.30 QMJHL 1.48  1.05 
Rob Schremp           2002-03 16.50 OHL   1.14  1.03 
Rick Nash             2000-01 16.54 OHL   1.14  1.02 
Jason Spezza          1999-00 16.55 OHL   1.17  1.01 
Mike Ricci            1987-88 16.18 OHL   1.49  1.00 
Pierre Turgeon        1985-86 16.34 QMJHL 1.65  0.99 

Most of these players did not approach Tavares' age-16 performance, but even at these lower levels, they averaged nearly a point-per-game in the NHL at age 18, and 10 out of 12 of them arguably ended up being NHL stars - while Martin Lapointe played nearly 1000 NHL games (and the jury is still out on Rob Schremp).

So what happened the next season? Tavares kept his output steady, which still placed him at the top of his age group:

Name                 Season   Age League PPG   Adj
Sidney Crosby        2004-05 17.40 QMJHL 2.71  2.55 
Jason Spezza         2000-01 17.55 OHL   2.10  1.89 
Pierre-Marc Bouchard 2001-02 17.68 QMJHL 2.03  1.74 
John Tavares         2007-08 17.28 OHL   2.00  1.74 
Pavel Brendl         1998-99 17.77 WHL   1.97  1.68 
Dale Hawerchuk       1980-81 17.74 QMJHL 2.54  1.62 
Ramzi Abid           1997-98 17.77 QMJHL 1.99  1.60 
Vincent Lecavalier   1997-98 17.70 QMJHL 1.98  1.60 
Marc Savard          1994-95 17.46 OHL   2.11  1.56 
Kyle Wellwood        2000-01 17.63 OHL   1.74  1.56 
Mario Lemieux        1982-83 17.24 QMJHL 2.79  1.55 
Joe Thornton         1996-97 17.50 OHL   2.07  1.55 
Pierre Turgeon       1986-87 17.34 QMJHL 2.66  1.55 
Rob Brown            1985-86 17.73 WHL   2.51  1.51 

A few more good names pop up on the list, along with some huge disappointments. My initial inclination was that because Tavares wasn't scoring at a higher rate, he wasn't getting any better. And players who don't get better as teenagers tend to be very disappointing professionals. But he's still at the top of every list at his age, and he doesn't play on a top offensive team like Crosby did. I think he's obviously not as good as Sidney Crosby, but likely better than Jason Spezza. That's not a terrible verdict - Spezza was 6th in NHL scoring last season; Crosby, whose abilities are well-known, was 1st in scoring two years ago.

The bottom line: Tavares probably won't be the best player in the game, but it's pretty likely that he'll be in the top ten even if he doesn't step up another notch in his final junior season.

Labels: Committed Indian, junior hockey, Projections

The 2008-09 KHL season is about 14 games in, and we can take our first stab at producing a league equivalency. There are 23 skaters in the KHL who played at least 20 games in the NHL last year. Their cumulative stats:

	GP	G	A	PP82
NHL 2007-08	1191	131	228	25
KHL 2008-09	260	54	76	41

That would give a league equivalency of 25/41 ~ 0.60. However, the KHL has much lower scoring and a lower assist-per-goal rate than the NHL, so we need to adjust this figure by the overall PPG scoring rates of the two leagues = 0.41 (NHL) / 0.36 (KHL). This gives a league equivalency of 0.70, which is somewhat lower than Russian leagues have had in the past, albeit with much larger numbers of players and games played in the sample.

Labels: Equivalencies, KHL, Projections

I now have five-to-seven years of player data for Canadian Junior A leagues, so it's now possible to generate equivalencies for each league. I ran the data for British Columbia (BCHL), Alberta (AJHL), Saskatchewan (SJHL), Manitoba (MJHL), Ontario (OPJHL), Quebec (QJAHL) and the Maritimes (MJAHL).

The Quebec and Maritimes leagues did not send a significant number of players to the NCAA, and few players went from the SJHL to Major Junior. The Manitoba league was more likely to send players to other Junior A league than to have them step up to a higher-level of hockey.

Here are the equivalencies normalized to a player age of 18:

Junior A to NCAA	EQUIV	Mean Age	N
*USHL	0.66	19.41	384
AJHL	0.39	18.38	91
BCHL	0.34	18.34	190
OPJHL	0.32	18.33	94
SJHL	0.31	18.37	42

A huge drop from the USHL to the Junior A leagues - but not a huge difference among the Canadian leagues. The AJHL seems to have a higher level of play.

Junior A to Major Junior	EQUIV	Mean Age	N
AJHL	0.61	17.43	48
BCHL	0.57	18.70	34
QJAHL	0.51	18.51	30
MJAHL	0.47	17.62	31
OPJHL	0.44	18.44	217

Less of a difference this time, but again the AJHL is on top.

Major Junior to Junior A	EQUIV	Mean Age	N
BCHL	0.33	18.37	33
SJHL	0.27	19.17	29
AJHL	0.22	18.80	43
QJAHL	0.21	18.71	85
MJAHL	0.20	19.23	19
OPJHL	0.17	18.20	71

Note that the equivalency is in this table is actually for Junior A to Major Junior so that we can compare the relative performance of players stepping up a level of hockey versus those stepping down. Junior A leagues appear substantially less difficult for players coming from Major Junior than they did to players going in the other direction.

Overall, assuming that the level of play is identical in all three Canadian major Leagues (which has been shown elsewhere on this site), then the BCHL and AJHL appear to have the highest level of play of all the Junior A leagues.

Labels: Equivalencies, junior hockey, Projections

"Can Leafs overcome the odds? Just when it seemed that any chance of reaching the post-season was lost, hope has been rekindled..."

Toronto Star, March 3rd, 2008 [link here]

Hope certainly springs eternal in Leafs land - I know you've got to believe. But when you talk about overcoming the odds, wouldn't it be good to know what the odds are?

I ran the rest of the schedule using a pretty simplistic model: I assumed that every team was as good as their pythagorean record suggested at this point. That's not strictly true, since injuries and trades will change a team's inherent ability...but I'm really interested in the quick-and-dirty answer here. I also assumed that the outcomes of shootouts and OT were completely random...because they certainly appear to be.

At any rate, here are the Leafs chances based on 10000 random trials:

In playoffs: 3.6% [Out of playoffs = 96.4%]

8th place = 2.6%
7th place = 0.9%
6th or better = 0.12%

Their average record going forward is 6-9-1, so it's not surprising they have really low odds to hit the 10-5-1 record posted by the average playoff-bound Leafs squad in my simulation.

My advice: don't try to budget your beer money to last into the playoffs...

Labels: Projections, Standings

Kyle Okposo made a few headlines this week: after the World Junior Championships are over, he's quitting the University of Minnesota and joining the New York Islanders. He was drafted 7th overall in 2006, which puts him a bit behind the curve in jumping to the NHL: he's the only player in the top 15 who hasn't played professional hockey. His former teammates Phil Kessel and Erik Johnson are playing regularly in the NHL.

So the question is, if the Islanders keep him in the NHL, what can we expect from Okposo for his next 40 games? This chart shows the ratio of NHL points-per-game to NCAA points-per-game for all players who've jumped from college to the NHL since 1979:

	10%	25%	50%	75%	90%
NCAA	0.15	0.21	0.29	0.39	0.48

So 50% of players retained 29% of their scoring in the NHL; just 10% retained 48% or more. Okposo has had a pedestrian 2007 NCAA season (11 points in 18 games), and if we combine that with 2006 (40 points in 40 games), his middle 50% scoring range amounts to 7-14 points in the second half of the 2007-08 season.

That seems a little conservative, and we should take into account that Okposo isn't 20 years old yet. This chart shows the difference in PPG ratios by age:



So we were probably 25% too pessimistic with our projection. Okposo is a good bet to pick up between 10 and 18 points in his next 40 games. That's low enough that the Islanders might just assign him to their AHL affiliate, and we won't see Okposo until next year.

Labels: Equivalencies, NCAA, Projections

"Getting to the big time early doesn't necessarily mean you've made it big. Or to stay." - Damien Cox, ESPN.com, 12/12/07 in Starting in NHL as a teenager doesn't mean instant success.

Sometimes you run across something that's so illogical, it must be true. And Damien Cox's article on ESPN was one of those jarring ideas: young players who make it to the NHL before their 20th birthdays aren't necessarily guaranteed of a good career - and it takes years to know whether they'll be any good.

Cox presents what seems like plausible evidence: looking at the 1999 season, not every player who made the NHL as a teenager turned into a superstar. While this group included Vincent Lecavalier, Simon Gagne and Scott Gomez, it also featured Manny Malhotra, Brad Stuart, Martin Skoula, Rico Fata, Mathieu Biron, Artem Chubarov, Steve McCarthy, Oleg Saprykin and David Tanabe.

So Cox concludes: "There's no discernible trend, other than to say, based on that group of players, being in the NHL by 18 or 19 is no guarantee of future success."

Well, maybe that's true, but let's extend our analysis beyond counting the number of "stars" in a single draft class and include every player who started his NHL career after 1967. Here we see the average career length based on a player's age on January 1st of his rookie NHL season:



The result is pretty clear: for every year older a player is when he debuts, he can expect to play 73 fewer games in his career. In other words, a 19-year-old rookie can expect to play until he's 29, and a 21-year-old rookie will also expect to play until he's 29. Starting late has no effect on when a player's career is over*.

But maybe longevity doesn't translate to scoring prowess? We can use career points-per-game (PPG) to estimate offensive skill:



Again, the younger a player is when he starts out, the higher his rate of offensive production.

Well, what about Cox's assertion that it takes years for teenage rookies to establish themselves as good player? We can look at the year-over-year change in PPG for these players:



Again, the general trend is higher as a player gets younger: older players improve their skills less from year-to-year than younger ones. But we already knew that. It's true of 20-year-old juniors vs 18-year-old juniors, so why wouldn't it be true of 18- and 20-year-old pros?

Unfortunately, to claim otherwise, we need to take a leap based on a small amount of non-quantitative evidence. While there's no guarantee that teenage rookies will turn into superstars, the younger a player is on average when he makes the bigs, the better his career will turn out to be. However many high draft pick busts there are, they don't outweigh the huge number of successful players who hit the NHL in their teens.


* - The youngest age group, which was no older than 18 years and 3 months when the season starts actually performs slightly poorer than the group that's six months older. This is because of small sample size: all other groups include 6 months of birthdays, while this group only includes three months of player since they needed to be 18 on the day the season started. This group also includes Sidney Crosby and Patrice Bergeron, who will most certainly skew this group upwards as their careers progress.

Labels: ESPN, Projections

The short column I wrote the level of play in the USHL has generated a lot of interest in the last couple of days.


Marc Foster's junior hockey blog made reference to it:

Junior Hockey Blog

And the USHL President's office and Director of Scouting got in touch with me to express their gratitude at seeing these numbers available somewhere. A big thanks to James Mirtle for his blog post on the USHL, which made me think it was finally time to analyze the USHL.

Labels: Equivalencies, junior hockey, Projections, USHL