|
Over the next little while in Up And Coming, we'll be using a database of OHL players to develop some projection methods slightly more rigorous than “Iain says so”. I compiled the statistics of every OHL forward first eligible for the Entry Draft between 1989 and 1998 that I could find, which ended up being 521 players. They range from Joe Thornton and Eric Lindros at the upper end to Rick Kowalsky and Ryan Stewart, who managed zero points in 47 games for the 1993-94 Windsor Spitfires. The OHL is the highest-quality Canadian major junior league, slightly ahead of the WHL and a good bit ahead of the QMJHL. Forwards are easier to project than defensemen or goalies, so we're starting there.
In addition to recording the players' junior stats, I compiled their stats for the 10 years after their first draft-eligible year, broken down into major league (NHL), high minors (AHL and the defunct IHL, as well as the better European elite leagues), low minors (second-tier European elite leagues, and the ECHL, CHL, UHL etc.), and junior (including Canadian university leagues).
The 10-year period is selected because this is typically the maximum number of years a team will have control over a drafted player. At age 27, if not sooner, the player will become an unrestricted free agent. How a player performs at age 29 is irrelevant to drafting decisions, since a player of that age will be a free agent. The minor league stats are included to give an idea of the shape of a player's career, as well as to recognize the fact that some players don't even get a chance to play in the NHL or not, regardless of skill or talent. These players will be evaluated based on their minor league stats, which of course are subject to a hefty discount compared to NHL games.
First, we need some sort of measure of a player's value. I won't spell out the specifics here, since the point is to develop a rough estimate rather than a true value. I combined a player's games and points, as well as points-per-game, at each level to arrive at a “Career” score. To illustrate that this does in fact identify the best players, below are the top 50 players from the database. Remember that Career rewards both longevity and production, and values NHL totals more highly than high minors, which are valued more highly than low minors, etc. Before dickering about any one player's placement, recall that only the first 10 years after the player's first draft-eligible year are covered, and this is not meant to be a definitive ranking of these players. It's just an estimate.
Rank Name Year GP Career
1 Thornton, Joe 1997 672 873
2 Lindros, Eric 1991 523 645
3 Arnott, Jason 1993 670 536
4 Nolan, Owen 1990 649 497
5 Allison, Jason 1993 623 472
6 Gratton, Chris 1993 770 435
7 O'Neill, Jeff 1994 731 432
8 Primeau, Keith 1990 668 409
9 Legwand, David 1998 610 398
10 Bertuzzi, Todd 1993 695 394
11 Ricci, Mike 1990 708 391
12 Savard, Marc 1995 617 355
13 Cheechoo, Jonathan 1998 660 333
14 Stillman, Cory 1992 689 322
15 Peca, Mike 1992 634 289
16 Dionne, Gilbert 1989 675 289
17 Fisher, Mike 1998 552 287
18 Bulis, Jan 1996 561 282
19 Murray, Glen 1991 680 281
20 Williams, Jason 1998 646 243
21 Cleary, Daniel 1997 655 229
22 Bell, Mark 1998 654 214
23 Avery, Sean 1998 622 200
24 Dawe, Jason 1991 661 193
25 Brunette, Andrew 1991 558 187
26 Wren, Bob 1993 651 182
27 Harvey, Todd 1993 626 181
28 Cooke, Matt 1996 592 181
29 Bowler, Bill 1993 646 178
30 Intranuovo, Ralph 1992 647 178
31 Smyth, Brad 1991 675 176
32 Moreau, Ethan 1994 737 176
33 McCarty, Darren 1990 591 174
34 Kostopoulos, Tom 1997 745 172
35 Ling, David 1993 733 170
36 Letowski, Trevor 1995 671 169
37 Skalde, Jarrod 1989 694 162
38 Malhotra, Manny 1998 622 159
39 Park, Richard 1994 753 158
40 Sim, Jonathan 1996 641 157
41 Brown, Kevin 1992 623 153
42 Taylor, Chris 1990 719 152
43 Craig, Mike 1989 587 151
44 May, Brad 1990 611 150
45 Sarno, Peter 1997 613 147
46 Willsie, Brian 1996 631 145
47 Firth, Jason 1989 662 143
48 McCauley, Alyn 1995 559 143
49 Shevalier, Jeff 1992 637 143
50 Mann, Cameron 1995 602 142
Joe Thornton tops the list, which should come as no surprise. Lindros is quite a bit behind, though that has as much to do with his injuries as anything. On a per-game basis, Lindros is quite close to Thornton. After that, with a Career score of 300 or more, we have a group of very good players, star players, but no one else who could be called a superstar. It's quite remarkable, really, the low number of really good players that came out of the league in this span of 10 years. Such players are truly rare, and as such identifying them and drafting them is of the utmost importance.
Since we have a Career value, we can evaluate how well NHL teams did in drafting these players. For a quick-and-dirty test, we can look at the coefficient of correlation between a player's draft position (with undrafted players given a value of 400, lower than any possible draft pick) and his Career score, which is -0.46. This is a pretty significant correlation (remember that a low draft position is good, and a high Career is good, producing the negative correlation). So it seems NHL teams do a pretty good job of identifying talent that will make an impact in the future.
However, the scouts have a competitor. It's a very basic statistic, very simple, which you wouldn't suspect would have such great predictive value given its straightforward nature. It's very simply points-per-game, adjusted for league scoring context (PPG+), in the player's draft year. Surprsingly, this number all by itself does a better job of predicting Career than draft position does; its coefficient of correlation with career is 0.68 (a high PPG+ correlates with a high Career). Without any other adjustments, this single stat seems to do a better job of projecting the success of OHL forwards than NHL scouts do as a whole. Some NHL teams may do better than others, but that is beyond the scope of this column.
Let's look at some numbers broken down by PPG+. I selected the following groupings of players to get a nicely-skewed distribution, which more or less fairly reflects the distribution of players' points-per-game averages in a typical hockey league. The following table shows the numbers of players in each group, the percentage who played professional hockey, the average Career, the average professional games played, and the percentage breakdown of pro games between major, high minor and low minor.
Group PPG+ Players % Pro Career Pro GP %Maj %High %Low
A 1.75 + 10 100 436 606 79 20 1
B 1.35 - 1.74 20 100 188 538 54 39 7
C 1.00 - 1.34 37 97 100 395 29 45 26
D .70 - .99 76 91 70 335 27 45 28
E .45 - .69 96 89 40 248 17 41 42
F .25 - .44 140 77 27 213 14 32 54
G .10 - .24 124 70 10 129 5 25 70
H .00 - .09 18 67 8 116 0 10 90
There are some nice clear patterns here. As PPG+ decrease, so does the percentage of players who play professionally, the average Career, the number of professional games, and the proportion of pro games played at higher levels. Group A is miles ahead of the rest; seeing as how it includes Thornton and Lindros, this is no surprise. It's also expected because the few best prospects tend to be substantially better than the second tier, which tends to bunch together more. I find the subtle differences between Groups C and D interesting: Group D's Career is a bit lower, though it's basically attributable to career length, which is about one season shorter. The proportion of games at each level is nearly identical.
We can look at a similar breakdown for players by draft position. We run into a problem with the distribution here, since 330 of the 521 players were never drafted. So we'll start at the top with about the same distribution, and then lump the remainder together when we have to.
Group Drafted Players %Pro Career Pro GP %Maj %High %Low
A1 1 – 4 12 100 339 562 87 13 0
B1 5 – 14 21 100 164 496 67 31 2
C1 15 – 44 35 97 123 415 48 33 19
D1 45 – 139 75 93 61 370 16 64 20
E1 140 + 48 98 38 271 11 42 47
F1 Undrafted 330 74 24 170 9 26 65
For a fair comparison, of course, we need to rejigger our original table, to make the distribution of players equal, like so:
Group PPG+ Players %Pro Career Pro GP %Maj %High %Low
A2 1.62 + 12 100 415 616 83 16 1
B2 1.26 – 1.61 21 100 156 476 46 46 8
C2 .99 – 1.25 35 97 103 410 27 44 29
D2 .70 - .98 75 91 70 335 27 45 28
E2 .56 - .69 48 88 50 279 20 42 38
F2 .55 - .55 330 76 20 177 11 31 58
While PPG+ does a better job of projecting Career, especially in Group A and except for the blip in Group C, there is one thing that draft position does predict much better: a high proportion of games played at the higher levels, among the higher draft positions at any rate. This makes a great deal of sense, given that the North American high minors over this time period were nealy all NHL farm teams. These teams are more likely to be stocked by draft picks rathe than free agents. A player with good skills but who was overlooked by the NHL was more likely to sign in the ECHL or somesuch league, since they often had little choice if they wanted to play professional hockey.
Next week, we'll look at this year's OHL draft crop, and use the database of players to identify some comparable players to see what this might reveal about their future prospects.
Iain Fyffe is an author of Puck Prospectus.
You can contact Iain by clicking here or click here to see Iain's other articles.
|
