October 5, 2009
Up and Coming
Projecting Defensemen

by Iain Fyffe

Welcome to the first regular installment of Up and Coming for the new season. I spent the summer working on a variety of projects, one of which was expanding my projection system to include CHL defensemen and goaltenders. This week we take a look at some of the results for defensemen.

There are a couple of adminstrative matters to get out of the way first. I have decided to change the name of the Career Score to the 10-Year Score (abbreviated 10Y), for the obvious reason that the score does not reflect a player's entire career, but rather the first 10 years after his first draft-eligible year. The Estimated 10-Year Score, which is the output of the projection system, is abbreviated 10YE. The tax accountant in me enjoys the alphanumeric aesthetics of these abbreviations.

I also felt that the system should've had a name by now. A clever bacronym of sorts seems to be de rigeur in this field. However, I prefer to use the suffix “-inator” whenever possible, so for now at least, we'll call it the Projectinator.

The development of the projections for defensemen followed that of forwards closely. (See here for the most recent discussion of forwards.) Of course alterations had to be made, both in calculating a fair 10Y value for defensemen, and in using junior statistics to project future success. 10Y scores for defensemen were designed to approximate the distribution of values for the forward scores, so that when it comes to combining the two groups they are both represented fairly.

Let's have a look at the top 20 defensemen by 10Y scores for the years in question (1989-1998, only players' first draft-eligible years), along with their estimated value. The Pro GP column represents the player's professional games played in the 10 seasons after his first draft-eligible year.

Player	           Pro GP  10Y	  10YE
Pronger, Chris	   642    1.03    1.00
Jovanovski, Ed	   643	   .92	   .63
Berard, Bryan	   510	   .91	   .81
Niedermayer, Scott 654	   .90	   .93
Morris, Derek	   564	   .89	   .53
Redden, Wade	   629	   .85	   .87
Sydor, Darryl	   623	   .80	   .81
Stuart, Brad	   579	   .79	   .76
Hatcher, Derian	   587	   .76	   .69
McCabe, Bryan	   633	   .76	   .75
Mara, Paul	   562	   .76	   .80
Sopel, Brent	   488	   .69	   .45
Brisebois, Patrice 498	   .69	   .84
McLaren, Kyle	   514	   .69	   .65
Boucher, Phillipe  447	   .65	   .62
Brewer, Eric	   543	   .65	   .42
Campbell, Brian	   549	   .65	   .47
Quint, Deron	   490	   .63	   .47
Boynton, Nick	   508	   .62	   .77
Skoula, Martin	   690	   .62	   .53

Overall I'm quite satisfied with the group of players we have at the end of this list, though I'm not overjoyous with the precise ordering at the very top. Clearly, Jovanovski and Berard do not belong ahead of Niedermayer. Pronger doesn't either, though, ultimately the precise ordering doesn't matter so long as you have the right group of players at each level of success. The 10Y scores are designed to allow the estimate to be built upon them, and not to rank current players. Ironically the 10YE scores do a better job of ranking the very top players, putting Niedermayer and Redden behind Pronger in the top three.

If you look at the top 20 in terms of 10YE instead of 10Y, you find David Cooper at .72 (compared to actual 10Y of .51), Nick Stajduhar at .69 (.39), John Slaney at .65 (.61), Richard Matvichuk at .65 (.54), Chris St. Croix at .65 (.28) and Chris Phillips at .64 (.60). Some of these make the system look bad, but some fooled the scouts as well. Cooper was drafted 11th overall in 1992, and Stajduhar was 16th in 1993. Slaney and Phillips have their scores pegged almost exactly by the projection, and Matvichuk's not far off. St. Croix is really the only blemish for the Projectinator in this group.

What goes into these scores? For the 10Y, a player's games played, points, and penalty minutes are combined, which includes all levels of play in the 10 years after the first draft-eligible year (major, high minor, low minor and amateur). The most interesting mechanical manipulation is with penalties. There is a sweet spot for defenseman penalties, both in play and in the projections. Too few penalties often indicates that the blueliner isn't putting the required minimum amount of physicality into his game, and like it or not, playing defense in the NHL requires a certain amount of physical play. Too many penalties, on the other hand, tends to show a player who takes unnecessary trips to the sin bin. Good penalties are taken in order to prevent good scoring chances and bad penalties are taken on plays where the skater could have prevented the scoring chance without taking the penalty. For instance, a scenario in which a poke check could have been as effective as a slash, or a legal body check could have replaced a hook.

Moving from the 10Y score to the 10YE, we obviously consider the player's scoring stats and penalties, but we also consider their age (month or birth), the league in which they play, and their height.

There is truth to the idea that the WHL produces better defensemen than the other major junior leagues. Actually, the WHL produces better forwards as well, but the degree to which the defensemen are better is greater. When calculating the 10YE for forwards, the OHL is weighted at 1.00, the QMJHL at 0.94 and the WHL at 1.10. For defensemen, the weights are OHL 1.00, QMJHL 0.90 and WHL 1.15.

Height is not irrelevant in predicting a junior player's future success. All else being equal (talent, work ethic, etc.) a larger player has an advantage in his 10Y over a smaller one, particularly when it comes to defensemen. Though, there are two very important caveats here:

• It is unclear how much of this advantage is genuine and how much of it is an illusion driven by the a priori assumption by hockey executives and coaches that size is important. There is no truly objective measure of a player's effectiveness yet, and part of a larger player's apparent effectiveness may be the result of receiving greater playing time simply because of his size.
• The degree of importance that a player's size has on projecting his future success is severely overestimated by the majority of professional scouts. Whereas a player's physical stature might properly downgrade his prospects from an A to an A-, some scouts take one look at a small player and slap a big, red TOO SMALL FOR THE NHL label on his brow, regardless of talent or other factors.

For further evidence of the size bias in scouting, we can look at all defensemen from the years in our database that were drafted in the first 20 spots in the draft, and then take the 12 of these men with the lowest 10Y score. Doing so gives us a group with something in common; can you guess what it is? If you said “they're all 6-foot-3 or taller”, you can move to the front of the Up and Coming class. Here is the list:

Player	          Year	  Drafted  Pro GP   10Y	  10YE	 Height
Belak, Wade	  1994    #12	   385	   .24   .47	 6-5
Soules, Jason	  1989	  #15	    52	   .25	 .27	 6-3
Brown, Brad	  1994	  #18	   483	   .27	 .33	 6-3
Bilodeau, Brent	  1991	  #17	   499	   .29	 .58	 6-4
Aitken, Johnathan 1996	   #8	   487	   .29	 .41	 6-4
Ware, Jeff	  1995	  #15	   312	   .30	 .31	 6-4
Larocque, Mario	  1996	  #16	   524	   .37	 .40	 6-3
Bennett, Adam	  1989	   #6	   192	   .38	 .44	 6-4
Stajduhar, Nick	  1993	  #16	   400	   .39	 .69	 6-3
Ference, Brad	  1997	  #10	   512	   .40	 .39	 6-3
Focht, Dan	  1996	  #11	   468	   .41	 .40	 6-6
Ward, Lance	  1996	  #10	   471	   .41	 .31	 6-3

Note that the Projectinator does a pretty good job of pegging these players' future values. There are a few defensemen overrated by the system (Belak, Bilodeau, Stajduhar), but of course they were overrated to a much greater degree by the scouts.

So who were the players most underrated by the scouts, according to their draft positions? We have Brent Sopel with a 10Y of .69, drafted at #144 in 1995. Brian Campbell (.65 10Y) went at #156 in 1997, while neither Andy Delmore (.61 10Y), Derrick Walser (.56 10Y) nor Jim Vandermeer (.55 10Y) were drafted in their first years of eligibility. However, the Projectinator wouldn't have liked these guys either; none have a 10YE of more than .47, so these skaters fooled both the scouts and the system. None are taller than 6-1, mind you.

There are also three other defensemen to consider: Darren Van Impe (.57 10Y, .55 10YE, 6-1), Andrew Ference (.56 10Y, .55 10YE, 5-10) and Chris Snell (.56 10Y, .51 10YE, 5-11). Ference was drafted at #208 in 1997, while the other two weren't even drafted in their first years of eligibility. These are definite wins for the Projectinator over the scouts, and again demonstrate the bias against smaller players.

Overall, the Projectinator hasn't found many overlooked gems among the defensemen. It does, however, steer you away from the bad early-round selections (generally hulks with little talent). Scouts do a pretty good job of identifying the good defensemen, but they also get a good number of false positives which eat up valuable early-round draft picks. Players like Johnathan Aitken should not have been drafted #8 overall, The Projectinator sees this, but unfortunately the scouts all too often do not.