Late last month, Jason Gregor looked back at the NHL draft from 1996 to 2005, in an attempt to put a value on NHL draft picks. Gregor’s far from the first to do so; after the jump I’ll look at his work and that of others to see what we can say about the NHL Draft.
Jason Gregor’s Take
Let’s start with Jason’s article. There are a lot of things to like about it; in particular the fact that the dataset is huge (10 years, ~2700 players) and the criteria is relatively straight-forward. That said, there are two things that I think could be improved: first, there is some subjectivity in the player ratings, and second Gregor breaks things down by round, meaning that we don’t know how much difference there is between a top-10 selection and a 10-20 selection. The latter is more of a problem than the former, as we will see later on.
It’s still quite a good article, and the key points I took from it were as follows:
- A first round selection is far more valuable than any other selection – a player taken in that range has a slightly better than 60% chance of turning out to be a “decent” or better player; nearly three times the rate of the second round.
- Second and third round picks are a cut above the rest of the draft, although the chances of landing a “decent” or better player are quite low: a little better than 1 in 5 for the second round, and slightly better than 1 in 7 for the third round.
- There’s little difference between picks made from the fourth to ninth rounds of the draft; over those rounds players are roughly half as likely to develop into a “decent” or better player as a third round pick.
Scott Cullen’s Take
Over at TSN, numbers guy Scott Cullen did something similar to Gregor’s article in mid-2009. Like Gregor, he used a 10-year period (1995 to 2004) and subjective rankings. However, Cullen broke the draft numbers down into groups smaller than full rounds – groups of five for the first 60 picks, and then groups of 15 afterwards. A few months later, Cullen clarified things even further for the first round – using 30 drafts and breaking down the top thirty picks by each draft spot.
The key points I took from Cullen’s work:
- There’s a huge gap between a top-three selection and a four through six selection, which is followed by another gap between the sixth spot and the rest of the draft.
- Again, there’s a big drop between the first round and the second/third rounds, and another big drop from about the 100th pick on, at which point things flatten out – there’s very little difference between drafting 105th and 195th.
- A player taken after the first round but in the first hundred picks has roughly a one-in-three to one-in-four shot at hitting the 100-game plateau in the NHL.
The Hockey Prospectus Take
Again from the summer of 2009, articles by two different writers at Hockey Prospectus. Richard Pollock considered both first round selections and late round selections, while Tom Awad considered the entire draft. Pollock uses both points and games played, while Awad uses Goals-Vs.-Threshold, a unified statistic that compares the value players of all positions generate above replacement-level players.
All the articles are good, but Awad’s method does the best job of comparing players across position objectively, and I like the fact that he weights his findings to reflect the fact that immediate production is better than later production. Plus, his fitted graph does the best job of showing how precipitously draft pick value drops off.
- The first few picks in the draft – especially those in the top five – are vastly superior to other picks, including picks later in the first round.
- Once again, picks outside the first round but in the top-100 are roughly equivalent in value – there is some value in having a higher pick, but that value is not especially strong.
- After the first 100 or so picks, there is almost no difference in the value of a draft pick.
Despite the different methods used and the various writers employing them, the findings from one article to another are extremely similar. I would argue the following, based on the data presented:
Get top-five picks. The top picks in the first round are very, very valuable. Teams moving up into this range cheaply are getting real value. Generally it costs an arm and a leg to acquire a marquee pick, but there are exceptions – for instance, Tampa Bay’s trade of the fourth overall pick in 2002 to Philadelphia for Ruslan Fedotenko.
After the top-five, move down in the first round. Teams trade up all the time to snag players that they like in the first round, but as a rule it is a very bad idea. Teams trading down don’t really lose much value, especially if they aren’t trading down very far.
Top-100 picks outside the first round are not especially good, and are virtually interchangeable. These picks are all about quantity over quality; good players are there to be found but the odds are stacked against any individual selection. Trading a mid-second round pick for two third round selections would seem to be a wise move the majority of the time.
Picks outside the top-100 are all but interchangeable. These picks are all about quantity. Every selection is a long-shot, teams play hunches and snag the favourites of regional scouts, and basically there is very little to separate the players involved except the biases of individual teams and who their scouts have seen.
Contending teams can trade mid- to late-round picks with minimal reluctance. Draft picks after the first round, and especially those outside the top-100, can be spent with almost no hesitation on actual players. For a contending team, acquiring depth or quality via draft picks is an extremely sound strategy.