clock menu more-arrow no yes mobile

Filed under:

NHL Drafts: A Quantiative and Historical Comparison

New, 1 comment
At the trade deadline this spring I did a short series of posts where I went back and looked at how the Atlanta Thrashers did compared to all the other NHL teams in the 1999-2002 Entry Drafts. I wanted to do a more extensive analysis later in the year. Now that the off-season has arrived and the 2007 Draft is less than two weeks away I will return to this topic.

With this series I want to answer several questions for each draft between 1980-2004.

  • Which teams did the best at finding valuable NHL players?
  • Which teams did the best they could given where they picked in the draft?

How Much is That Player Worth?
In the review of each NHL draft, first I will rank each team in terms of the total NHL value of the players they selected.

What is the best way to estimate a player's value? My first thought was to use NHL games played. Games played is a handy measuring stick of value. A player is in the line-up because the GM or Coach thinks he brings more to the table than other players available within the organization. Using games played also gives credit to defensive defensemen who may be undervalued (someone like a Ken Daneyko) if you simply look at points.

As much as I like games played, it simply doesn't work well enough. For example, there are third/fourth line guys (like Rob Ray for example) who dress for many NHL games but rarely see the ice or rarely impact the outcome of the game. Therefore, I have decided to go with Tom Awad's Goals Versus Threshold (GVT) player valuation system because I think it is a significant improvement over using games played. The GVT system divides credit for goals scored and goals prevented among the players on a roster for all NHL seasons after 1950.

No player valuation system is perfect, but I think that this about as good as you can do with the data available at to researchers . The GVT system also very useful because Tom has produced values for all NHL players since 1950 which is necessary for a project like this one that seeks to
compare NHL drafts over time.

(Note: Because the valuation system does produce negative values for players below replacement level I have made a small upward adjustment on values in ensure that all skaters end up with positive values. Why? Because even a bad 4th line player who dresses for 10 games is still better than a draft pick who never even plays a single NHL game. Therefore draft picks who get a cup of coffee at the NHL level need to be valued higher than busts who never dress for a single game.)

Did the Team Pick the Best Available Player?
One of the really difficult problems in comparing one NHL team to another is that every team picks at a different point in the NHL draft order. The teams that pick first obviously have better options available to them than the the teams that pick later in the order. How do you control for that?

What I have done here is create an opportunity cost batting average for each team. The basic idea is simple enough. Did a team pick the best player available to them when it was their turn to pick? If the answer is yes, they batted 1.000 if the answer is "no" then they missed out on receiving the maximum value available to them at that point in the draft and therefore they should a score of less than 1.000 (The difference between 1.000 and something lower is the opportunity cost of passing on a the best available player).

Now if a team passes on a great player and instead drafts a complete bust they receive a score of zero. On the other hand, if a team passes on the "best" players available but still drafts a guy who makes the NHL, then we have to give them some credit for picking a NHLer. The amount of credit they receive depends on how difference in the value between the guy they took and the guy they passed up.

For example, let's take a look at the first four picks of the 1983 NHL Draft. All four picks at the top of the draft became NHL players but there is a huge difference in terms of the value of their respective NHL careers. The best player available was Steve Yerman who was taken 4th overall. Thus the Red Wings would receive credit for a perfect selection (1.00). The scores for the teams that passed on Yzerman depend upon the value of the player they took instead of Yzerman. The NY Islanders took Pat LaFontaine who had a great career cut short by injury. His career value (446) was worth roughly half the value of Yzerman's (823) and thus the Islanders received a score of .54 since that is the LaFontains's value (446) divided by Yzerman's career value (823). In selecting LaFontaine over Yzerman the Islanders obtained a very good NHL player but they missed out on maximizing their potential return on their draft pick and thus do not receive a perfect score.

1983 NHL Draft First Four Picks
Pick Player NHL Value Opportunity Cost Value
1 Lawton 165 0.20
2 S. Turgeon 265 0.32
3 LaFontaine 446 0.54
4 Yzerman 823 1.00

Each team receives an opportunity cost "batting average" for each draft. If a team took the most valuable player left in the draft every time they picked, that team would have an average of 1.000 which is the theoretical maximum. In reality the NHL Draft is very difficult. In baseball a .300 batting average is thought of as a very good number for a hitter. In the NHL Draft the average NHL teams only scores about .100 in a typical draft. A "good" draft batting average is .150 and anything over .200 is a "great" average. Success in the NHL draft is terribly difficult to achieve. The failure rate of NHL picks is absolutely staggering when you look at it.

So the following series will provide two different takes on recent NHL Draft history. First it will allow us to judge which teams got the most total value out of their picks. Second it will allow us to see which teams most consistently get the maximum amount of value out of the draft regardless of where they are in the draft order.