Brian Burke hasn't always looked like a genius in the NHL draft. But the current general manager of the Toronto Maple Leafs has made two of the most bold and memorable draft-day moves in the history of the game.
In 1993, as GM of the Hartford Whalers, he swapped a handful of draft picks for the second overall choice and selected Chris Pronger. A decade ago, facing what has emerged as one of the weakest draft pools ever, Burke bid his way to the second and third picks and then drew a pair of Sedins.
Squarely in his crosshairs today is John Tavares. While it takes two (or more) parties to make a successful swap happen, there is considerable evidence that the man just might pull it off.
Why this matters to the Leafs is not just that Tavares is considered by Burke to be the top talent available in the draft pool but that such a selection would be the closest thing to a sure thing that one can get in the draft.
So this scenario is about risk management as much as talent management. The lottery to order the top picks of the NHL draft is high profile but the real lottery is in the assessment and emergence of talent from the actual draft.
What is clear to most is that the top selections offer the best chance to acquire an NHL-calibre player who will perform at a high level. What is less clear is how those odds erode as the draft progresses.
To assess that question let's look at the 20 drafts between 1985 and 2004 to see how careers have developed. Later draft years were excluded as many players graduating from the process have yet to fully mature. Also excluded from this analysis, for comparative reasons, are goaltenders. Note that many of the players included in the study have far from complete careers.
Of the top 30 picks from each of the drafts over the study period around 10 per cent were complete duds, failing to play even one game in the NHL.
If we expand the definition of a first-round dud to playing fewer than 200 NHL career games (perhaps an unfair definition for those drafted in the later years of this study), then more than 25 per cent would qualify. Looking at this by draft position we can count nine "failed'' draft picks among the top five choices (no 'duds' were drafted No. 1 overall), 24 from picks 6-10, 32 from picks 11-15, 40 from picks 16-20, 33 from picks 21-25 and 38 from picks 26-30.
Does this not provide solid support for Mr. Burke's behaviour?
The top five overall picks, especially the first pick, offer clearly reduced risk profiles. After that the first round looks a great deal like a lottery (although picks 6-10 look somewhat safer).
If you get to roll the dice often enough, a luxury many GMs don't have, then you can think about averages more than absolutes. Below is a graph of the average performance (games played and scoring points) of forwards selected among the top 30 in the draft.
Forwards drafted No. 1 overall have averaged over 800 career games and 700 career points. Those averages fall off rapidly over the first few picks and diminish slowly as we go deeper into the first round.
Such an analysis suggests that, after the top 10 selections, there is not much difference in the value of a first-round draft pick. This is more confirmation that, beyond the blue-chip prospects, even the first round of the draft should be thought of as a lottery.
Defencemen need to be assessed separately as they are not properly measured by a focus on scoring points. While overall impact can be measured using complex methods, the simplest way to look at the impact of defenders is to look at ice time. But the NHL has captured that statistic only recently in its history. The next best proxy is games played.
Below is a graph of the average games played (and points) of defencemen selected among the top 30.
Because the dataset is smaller, this is a noisier-looking graph. But the games played profile looks similar and the conclusions reached for forwards would also seem to apply to blueliners.
If the latter part of the first round is a lottery, what does that say about later rounds?
Over the study period the NHL grew from 21 to 30 teams. To study the value of a later-round pick we need to normalize for this. Let's group the first 30 draft picks as "Tier 1'', the second 30 picks as "Tier 2'' and so on. And let's chop the data off after 240 picks (Tier 8) due to variations in the length of the draft. The data already reviewed suggests that it might also be helpful to split Tier 1 into the top 10 picks (Tier 1a) and the subsequent 20 picks (Tier 1b).
Below is a graph of player performance by draft tier.
Intuition tells us that the value of a draft pick ought to diminish with picks that are deeper in the draft. Sure enough, Tier 1 picks outperform those in later tiers. But this analysis also suggests that there is not much difference in expected player performance between the subsequent rounds of the NHL draft.
That there is no material difference in the value of a draft pick after the first round is further strong evidence that the draft is largely a lottery. Some very talented players, such as Theo Fleury, Peter Bondra, Pavol Demitra, Tomas Kaberle, Pavel Datsyuk and Henrik Zetterberg have emerged from quite deep in the draft. And this assessment has ignored goaltenders. A large number of today's No. 1 goaltenders were selected with a low pick.
How does one increase the chances of winning a lottery? Hold more tickets. Want to reduce the overall risk in draft results? Make more picks.
The selection of a blue-chip prospect substantially reduces draft execution and concentrates talent in one player - both good outcomes. But the relative value of late picks is higher than one might think. To build a successful team over time one needs to believe in these dynamics.
Ken Holland has built the Red Wings into a powerhouse by, among other things, successfully managing this lottery. No other general manager has done so much with so little.
In Toronto the story has been different. The Leafs have historically been short on draft picks. In particular, Cliff Fletcher did not apply this data as the interim GM in Toronto. As he was cleaning house he traded away more picks than he acquired.
Burke's pursuit of John Tavares suggests that he gets it. In this case the improved certainty of outcome is as valuable as the improvement in talent.