March 18, 2009
Howe and Why
Career Projections for Rookies

by Robert Vollman

In order to gain insight into a particular player, we look for players with similar statistics throughout the NHL's 90 year history. The more data we have, the more valuable those insights are likely to be. Ironically, the less information you have about a player, the more you'll want that insight.

Let's take rookies, for example. As we explored in my inaugural article almost a month ago, looking for similar players based on a single season isn't bound to provide very useful insights, because the sample size is just too small. Players can have exceptionally good or bad seasons for any number of reasons, like injuries, or particularly skilled or unskilled linemates, teams or coaches.

As a demonstrative example of the futility of looking for similar players based only on a single rookie season, consider Joe Juneau. Juneau had a tremendous rookie season, scoring 32 goals and setting a record with 70 assists with the Bruins in 1993-94. Based on that, a search for similar players would have yielded Joe Sakic, Teemu Selanne, and Mike Rogers. However, Juneau never again scored 20 goals nor more than 28* assists.

Let's look at David Perron's rookie season last year with the St. Louis Blues where he scored 13 goals and 14 points in 62 games, and see how insightful the results will be. Here are the ten closest matches, and how well they scored in their following season.

Player            GP  G  A PTS PIM

Lanny McDonald    64 17 27  44  86
Stan Gilbertson   66  6 15  21  19
Milan Michalek    78 26 40  66  36
Chris Oddleifson  60 16 35  51  54
Percy Galbraith   42  6  5  11  26
Bronco Horvath    DID NOT PLAY
Brandon Bochenski 41 13 11  24  16
Mike Grier        66  9  6  15  73
Magnus Arvedson   80 21 26  47  50
Marian Hossa      78 29 27  56  32

While there is certainly some insight to be gained from this list, there certainly aren't any obvious correlations between these players. Lanny McDonald, for instance, has about as much in common with Percy Galbraith as Ronald McDonald. Short of calling Iain Fyffe, how can we sort through this list and determine which young rookies genuinely compare to David Perron?

First, we're going to need more information about Perron. We'll take his previous season in the Quebec Major Junior Hockey League (QMJHL) and convert it to an NHL equivalent using the same techniques we used two weeks ago when studying the Russian Elite League. If we divide Perron's 39 goals and 44 assists in 70 games with the Lewiston MAINEiacs (I love that name!) by 4, and his 75 penalty minutes by 3, we can estimate that he would have scored 10 goals, 11 assists for 21 points, and received 25 penalty minutes if he had entered the NHL a year earlier.

Second, let's look for similar players using both this converted QMJHL season, and his actual rookie season, and see what new insights we can gain. Here are the new 10 closest matches, and how they did the following season.

Player           GP  G  A PTS PIM

Mark Freer        2  0  0   0   4
Dan Fritsche     69 10 12  22  22
Glen Murray      35  5  2   7  46
Steve Heinze     76 15 12  28  43
Danny Lawson     33  1  5   6   2
Matthew Stajan   82 10 29  39  44
Danton Cole      67 12 25  27  23
Antoine Vermette 77 19 20  39  52
Bill McCreary    73 15 17  32  16
Fredrik Modin    80 22 26  48  18

While there is still a disparity of styles and talent, this list of ten players is definitely more similar than the first. With only a few low outliers, these players scored between 0.4 and 0.6 points per game in the season following the match. The extra season certainly provided more insight into the types of players that historically scored similarly to Perron's first two seasons. At the time of writing, Perron is scoring right at the upper limit of that second list, though as more of a playmaker than a goal scorer. This second list gave us a bit more insight into what to expect from David Perron than the first.

The more information you have on which to base a comparison, the more accurate the matches will be. That's why it's generally not as useful to attempt career projections on rookies using Similarity Scores. Translating their statistics from other leagues gets you just enough information to start seeing the various potential avenues a player's career might follow, but don't forget to refine your projections after a few more seasons when it becomes more obvious which ones are most likely.

*[ed. - minor error corrected]

Robert Vollman is an author of Hockey Prospectus. You can contact Robert by clicking here or click here to see Robert's other articles.