Hockey Analysis for the Thinking Fan
 

Projecting Junior Hockey Players and Translating Performance to the NHL

How difficult is it to score a goal in the National Hockey League relative to another league? With half of NHL players coming from the minor leagues, a quarter from European Elite Leagues, 20% coming directly from Canadian Major Junior and 10% from the NCAA, that’s a question NHL teams try to answer every day. In evaluating these players, it is critical to know how a player’s performance translates to the NHL.

One way to evaluate the difficulty of one league relative to another is examine the relative performance of players who have played in both leagues. Players rarely play significant time in two leagues in the same year, but they often play in one league in one year and in another the next. As long as a player’s skill level is approximately constant over this two year period, the ratio of his performance in each league can be used to estimate the relative difficulty of the two leagues.

Historically, the only statistics available are goals, assists and games played. With so little data, the best quantity to compare is a player’s Point-Per-Game rate (PPG). The difficulty of a league relative to the NHL can be determined by dividing the PPG that a player had in that league in one year by the PPG he had in the NHL the next year, or vice-versa. The PPG should be adjusted for the assist per goal rate in each league since European leagues don’t award a second assist as often as the NHL does.

Minor Leagues

NHL teams have traditionally had minor-league affiliates in the American Hockey League (AHL) and the International Hockey League (IHL). The IHL was absorbed into the AHL in 2002, and all AHL teams are now affiliated with NHL teams.

LeagueDifficultyN
AHL0.44384
IHL0.43113

This method tends to underestimate the difficulty of leagues that are substantially weaker than the NHL. This happens because only the top players get called up to the NHL – before the call-up, they were on the first line and playing the power play; after the call-up, they’re 3rd or 4th liners. This cuts down on both their overall ice time, and on their power play time, which is when they’d get the best opportunities to score.

We can improve this estimate by considering only even-strength goals:

71 PlayersPPGESG/G
AHL 2002-030.830.20
NHL 2003-040.340.11
Ratio0.410.54

European Elite Leagues

European players did not enter the NHL on a large-scale until the early 1990s:

LeagueDifficultyN
Russian Elite League0.83101
Swedish Elite League0.7877
Czech Republic League0.7453
Finland SM-Liiga0.5476
Deutsche Eishockey League0.5274
Switzerland National League0.4330

These results are not unexpected given the performance of each country’s national team in the Canada Cup and World Cup of Hockey over the last thirty years:
TeamWPCT
Russia/CIS/USSR0.649
Sweden0.540
Czech Republic0.478
Finland0.305
Germany0.144
Canada0.681
United States0.562

Given that the best team outside of North America (Russia/USSR) has a lower winning percentage than the Canadian National Team, it is not surprising that the Russian Elite League, where most players are Russian, has a lower league difficulty than the NHL, where most of the players have historically been Canadian, and which now selects players from throughout the world.

Major Junior and College Leagues


Most young players don’t jump directly from Major Junior hockey to the NHL, and instead spend a year or more in the minors. The league difficulties for the top three Junior leagues are shown below:
WHL N OHLN QMJHLN
Junior to AHL 0.433020.45295 0.41135
Junior to NHL 0.301430.30205 0.2862
Implied AHL to NHL 0.70 0.67 0.68
Observed AHL to NHL0.65154

Only 18- and 19-year-old players were included in the AHL to NHL analysis. Note that the AHL league difficulty (~0.68) is higher for 18- and 19-year-olds than it is for players of all ages. This is because young players tend to improve substantially from one year to the next, while older players have reached their peak level of ability. Young players, as a group, will do much better in their next year in the NHL, which makes the AHL appear better than it actually is.

The difficulty of all three junior leagues is about the same. Also, the difficulty of each junior league with respect to the NHL (~0.30) is the same as the difficulty experienced by players who went from junior to the NHL via the minors, which validates the concept.

The next largest source of NHL players is the NCAA:

 

Difficulty

N

NCAA

0.41

295


Since NCAA players lose college eligibility if they declare for the NHL Draft, they do not typically opt-in to the draft until age 21 or 22. The NCAA league difficulty is not then affected by the skills improvement seen in teenage players.

The World Hockey Association


The World Hockey Association (WHA) was a North American professional league that operated from 1972 to 1979. Initially it competed against the NHL for the best players, but it ultimately went bankrupt and its remaining teams were absorbed into the NHL. The WHA league difficulty is shown below:

Year

72-73

73-74

74-75

75-76

76-77

77-78

78-79

WHA

0.46

0.76

0.70

0.88

0.55

0.65

0.89

N

39

20

21

16

10

14

59


In its first season, 1972-73, the WHA league difficulty was 0.46, which is barely better than the minors. This is not surprising since 39 former NHL players played in the 12-team WHA in its inaugural season, while 200 minor league players filled out the rosters.

During 1973-75, the WHA stole away top NHL draft picks, recruited more established NHL players, and signed skilled European players. The WHA league difficulty ranged from 0.7 to 0.88 during those three years. However, beginning in 1976, the balance swung back in the NHL’s direction, and the league difficulty suffered as a result. In its final year of operation, the WHA was almost as good as the NHL – 59 players moved to the NHL after the last season, and their performance implied a WHA league difficulty of 0.89.

Summary

The league difficulty for the largest NHL feeder leagues is shown below:

League

Difficulty

N

NHL

1.00

 

WHA Final Year (1978-79)

0.89

59

Russian Elite League

0.83

101

Swedish Elite League

0.78

77

Czech Republic League

0.74

53

Finland SM-Liiga

0.54

76

Deutsche Eishockey League

0.52

74

WHA First Year (1972-73)

0.46

39

AHL

0.44

384

IHL

0.43

113

Switzerland National League

0.44

30

NCAA

0.41

295

Canadian Major Junior

0.29*

410


* The difficulty of Canadian Major Junior is lower than predicted by this method since major junior players are 20 years old or younger and consequently experience significant skills growth from year-to-year.

Further Junior Hockey Projections

 

Twenty years ago, almost all every NHL player was drafted from the Canadian Hockey League. Even today, the majority of NHL draft picks still come from the three Tier I junior hockey leagues, the Western Hockey League (WHL), the Ontario Hockey League (OHL) and the Quebec Major Junior Hockey League (QMJHL). But how does player performance translate from these leagues to the NHL?

 

It is obvious that, despite drafting thousands of times, NHL scouting hasn’t adequately answered this question. Looking at the 14th-23rd post-expansion drafts, from 1979 to 1989, 15 of 33 picks had careers that lasted more than 1000 games. But NHL teams also used the second and third picks on players like Dave Chyzowski and Neil Brady, not to mention famous “busts” Doug Wickenheiser, Doug Smith and Perry Turnbull. Even after 20+ years of evaluating CHL players, some teams looked at Vincent Damphousse and Adam Foote and decided they’d be better off with Brady and Chyzowski.

 

So were these errors easily avoidable? That’s the question we’ll answer.

 

Junior League Parity

 

Some preliminaries: in the aggregate, there’s no significant difference between the three junior hockey leagues. This table shows the League Difficulty of Juniors relative to the AHL and NHL:

 

 

WHL

N

OHL

N

QMJHL

N

To AHL

0.43

302

0.45

295

0.41

135

To NHL

0.30

143

0.30

205

0.28

62

 

The League Difficulty is the ratio of a player’s points-per-game (PPG) in junior to his PPG in the next season in the AHL or NHL. While individual performances vary, the average player who moved from the OHL to the minors or the NHL played as well as the average player who started out in the WHL or QMJHL.

 

The significance of Age

 

Does it matter how old a player is when he puts up big numbers in Junior? Obviously it does – Wayne Gretzky had 70 goals and 112 assists in 64 games for the Sault Ste Marie Greyhounds as a 17-year-old in 1977-78. Seven years later, Dan Hodgson had the exact same statistics when he was a 20-year-old playing for the Prince Albert Raiders. Hodgson was drafted 83rd overall despite his prolific scoring, and had 74 points in a 114-game NHL career. (He is still active in the Swiss National League.) At age 22, Hodgson had 57 NHL points; Gretzky had already scored 1024 points between the NHL and WHA.

 

So in a qualitative sense, it’s obvious in this case that a 17-year-old player’s performance predicts a much better career than a 20-year-old’s stats. But there is also a strong quantitative relationship between past and future performance. Based on the performance of thousands of drafted players, we can predict how many points a player will score in the NHL when he’s 21-years-old. If he’s 17, four years later, we expect him to score at 72% of his junior rate. But if he’s 20, on average, he’ll retain just 26% of his scoring.

 

There is a caveat: younger players are a bit less predictable than older players. For a 17-year-old, the middle 50% range of the projection is from 45% to 98%, while for a 20-year-old, it’s from 17% to 33%. This wide range reflects how unpredictable future performance is for NHL players. From age 21 to age 25, Wayne Gretzky scored between 196 and 215 points each season, which is only a 10% variation, while this method predicts a possible 2:1 variation in scoring. The performance of an individual player is much more consistent than it is for the large group of drafted NHL players.

 

We could narrow the bounds of the projection if we had more data about the players. This method tries to capture a player’s performance despite having no information about linemates, ice time, injury status, size and performance in other seasons. Who you play with can have a profound effect on your performance: Rob Brown played with Mario Lemieux and had 49 goals and 115 points. The Penguins traded him away two years later, and without Lemieux setting him up, he couldn’t crack an NHL roster.

 

PPG Projections by Age

 

This chart shows the ratio of NHL PPG to Junior PPG for forwards at each age. The age is as of January 1st of a given season, i.e. any player born in 1961 is considered 18 years old during the 1979-80 season.

 

In the aggregate, players reach their peak performance level at age 22 and hold it for several years. What’s most significant about this chart is what it implies about the age at which a junior player posts a particular PPG. A 17-year-old player with 2 PPG in Junior can expect, on average, to score 1.5 PPG in the NHL at age 22, while an 18-year-old Junior doing the same thing has an NHL projection of 1.0 PPG, which is 40 fewer points over the course of a season. This is the difference between elite players (Joe Sakic, Denis Savard, Dale Hawerchuk) and much lesser players (Jimmy Carson, Terry Yake, Mike Bullard.)

 

This is very significant for the NHL Entry Draft. An entire year’s worth of players become eligible for the draft, but the players born earlier in the year have a peak value 35% lower than players born late in the year. This is obvious when you consider the difference that one year of physical maturity can make at age 17. In evaluating a player, it is critical to keep in mind his exact age, down to his month of birth.

 

Back to the Beginning

 

Let’s look at a sample of star players and busts drafted with the first few picks:

 

Player

PPG Age 18

PPG Age 17

Pick

Dave Chyzowski

1.53

0.50

#2 – 1989

Neil Brady

1.13

 

#3 – 1986

Perry Turnbull

1.11

0.97

#2 – 1979

Dale Hawerchuk

 

2.54

#1 – 1981

Denis Savard

2.51

2.26

#3 – 1980

Mario Lemieux

4.03

2.79

#1 – 1984

 

In Junior, Chyzowski, Brady and Turnbull didn’t perform nearly as well as Hawerchuk, Savard and Lemieux, and this difference carried over to the NHL. In the 1979 draft, teams had the option of drafting young players from Junior or from the WHA, which had signed away numerous underage juniors in 1978. The WHA was nearly as good as the NHL at that point (see: League Translations) while the WHL (where Turnbull played) was only about one-third as good. The Hockey News rated Turnbull as a top prospect, while ignoring former WHA players in their scouting.

 

Sometimes, the differences can be more subtle:

 

Player

PPG Age 18

PPG Age 17

Pick

Doug Wickenheiser

2.39

1.38

#1 – 1980

Vincent Damphousse

2.25

1.51

#6 – 1986

 

Their pre-draft statistics look almost identical, but Damphousse was born in December while Wickenheiser was born in March, making him nine months older than Damphousse during the season. Damphousse’s 2.25 PPG was accomplished largely at the same age as Wickenheiser had 1.38 PPG – a huge difference! The key point is that just because two players are eligible for the draft in the same year, it doesn’t mean drafting teams can ignore age differences between the two players.

The USHL: North America’s Fourth Junior Hockey League

 

The United States Hockey League (USHL) is the only Tier I Junior Hockey League operating entirely in the United States. Unlike the three junior leagues that make up the Canadian Hockey League, the USHL adheres to NCAA eligibility standards, which allows its players to continue on to college hockey before potentially moving on to professional careers.

 

Fifteen years ago, the USHL did not measure up to Tier I standards: it featured a lot of 9-6 games, and relatively few of its players continued their hockey careers along anything resembling a path to the NHL. But as of 2006-07, nearly 10% of NHL players had also played in the USHL:

 

 

This is not surprising given that the number of USHL players moving directly to NCAA college hockey has been steadily increasing over the same time period:

 

 

At the same time, the caliber of USHL players relative to the NCAA has jumped substantially. The plot below shows the relative points-per-game scored by USHL players in their first NCAA season. The red line shows performance adjusted for the scoring levels in each league; blue is unadjusted.

 

 

As you can see, a point in the USHL has become 40-60% more valuable over the last fifteen years, assuming that the level of play in the NCAA has remained constant. The translation from the NCAA to the AHL has been approximately constant during this era, so this is a safe assumption. The average age of USHL players has also remained constant, which is especially significant given the dependence translations have on teenage players.

 

It is difficult to compare the USHL and Canadian Junior Leagues directly because the groups of players do not interact. Few USHL players move to Canadian Junior (though they typically maintain their level of production), and virtually none move from Canadian Junior to the NCAA. The table below shows the translations between leagues that commonly feed each other:

 

 

NCAA

AHL

CHL

N/A

0.43

USHL

0.65

N/A

NCAA

------- 

0.53

 

We can then estimate the translation from the USHL to the AHL by multiplying the translation from the USHL to the NCAA by the translation from the NCAA to the AHL:

 

 Implied Translation

AHL

USHL

0.37

CHL

0.43

 

The accuracy of the previous step is affected by a typical two- or three-year gap between a player’s last year in the USHL and first year in the AHL. We can compare this to the observed translation from the USHL to the CHL for the relatively small number of players who played in both leagues:

 

 

CHL

USHL - Implied

0.86

USHL - Observed

1.12

 

It is clear that regardless of the analysis method, players who play in the USHL and CHL perform approximately the same once they get to higher-level leagues. Allowing for a year's worth of improvement, 18-year-olds in the USHL would maintain their level of scoring at age 19 if they transferred to Canadian Tier I Junior. Were they to transfer during the same season, they would likely maintain at least 80% of their scoring.

NAHL and Tier II Junior

It is important to keep in mind the gap between the USHL and non-Tier I junior hockey leagues.

Leagueto CHLmean Ageto NCAAmean Age
Tier II0.4517.30.3319.4
NAHL1.00*16.90.5019.6
* = small sample size

The USHL is clearly better than Tier II Junior Hockey.