Does the NHL draft system reward failure?
Hockey fans worry about the least little thing becoming "a problem" in the league or the sport. And they constantly propose solutions to these perceived problems without ever checking to see if the problems exist in the first place. The idea that losing is rewarded by the draft system is one of those situations. A great deal of digital ink has been spilled bemoaning the draft system and its consequences.
Simply put - the NHL draft lottery in its current state heavily rewards failure....Although I would not accuse any franchise of tanking a season in an effort to obtain a better draft slot, there is no doubting the temptation is a clear and present danger to the integrity of the game. [emphasis added]
--Christopher Ralph, Fixing the NHL Entry Draft Lottery System: Playoff For Number One, The Hockey Writers, April 8, 2012
But, overall, it's evident that Gold are [sic] Williams are right: tanking to win a top draft pick is fairly widespread in pro sports.
Evidently, it can pay to lose and it can pay biggest to lose biggest, which is why many fans have been consoled by the "Fall for Hall" and "Fail for Nail" campaigns. [emphasis added]
--Edmonton Journal, How to prevent ‘Fall for Hall' and ‘Fail for Nail' campaigns, March 3, 2012.
As it is commonly described, the "draft problem" goes like this: 1. The NHL Entry Draft rewards losers with the best draft picks. 2. Teams thus have an incentive to "tank" for a couple of seasons to stock up on top-5 draft picks. 3. This has become a viable way to do business in the NHL, and GMs have an incentive to have a few bad seasons now to build winners down the road. The solution is always to fix the draft system to prevent almost-awful teams from deliberately playing really awful hockey to get good draft position.
The thing is that the presumption that the draft system rewards bad play now with Cup contenders down the line is just that: a presumption. So I asked a very simple question--are teams being rewarded for having bad seasons by being able to build winning teams? How often do teams get good by being bad to build up draft picks? And how long does that success last?
I took twenty years of NHL seasons (1991-2012)--other than the introduction of a draft lottery in 1994, there have been no major changes to the draft in that time--and I identified the 5 best and 5 worst point totals for each season. I also looked at whether teams moved around a lot, which would be a sign of league parity. Standard deviations ranged from a low of 2.16 (Detroit) to 9.64 (Pittsburgh), but the vast majority of teams had standard deviations above 6 (23 teams). So most teams are moving around a bit rather than staying in one general area of the standings.
I then identified what constituted a very bad stretch of seasons, seasons that could reasonably be presumed to garner a number of quality draft picks and thus reward bad play the most strongly. A bad stretch is one that:
- Lasts at least 3 consecutive years
- With at least 50% of those years landing the team in the bottom 5
- And none of those years seeing the team more than 2 spots out of the bottom 5 (remember, we're testing for tanking, not for disappointment.)
I also identified what would be the reward: a stretch of very good seasons. It's a mirror of the bad stretch:
- Lasts at least 3 consecutive years
- With at least 50% of those years landing the team in the top 5
- And none of those years seeing the team more than 2 spots out of the top 5 (i.e., that team was dominant in the league for that span)
I then considered timing. Top 5 draft picks are expected to start making a difference rather quickly. Assuming it takes 2 years to get from the draft to the NHL (not true for every first round pick, but often true for top-5 picks), the player ought to be making a difference within the span of his ELC (four years or less). Six years is a long time to wait for a reward for bad behavior, but it could happen. So I identified the instances where a bad stretch was followed by a good stretch by anything up to 6 years.
I did not consider single bad seasons or good seasons for the simple reason that a good or bad year in the middle of a stretch of mediocre years can be attributable to any number of causes--Injuries, luck, divisional strength, timing of contracts. It's impossible to link single bad years to a conscious strategy of tanking or single good years to the effect of a draft pick. Further, my main question is whether there are, in fact, strong institutional incentives to play bad as a conscious strategy. A single season of success is less of an incentive than a series of opportunities at the Cup, and a single season of failure cannot be determined to be a conscious strategy.
I found 19 bad stretches experienced by 12 different teams, and 11 good stretches by 7 different teams. So right off the bat, it's roughly twice as easy to be bad than to be good. [By the way, the Detroit Red Wings are unquestionably the best franchise of the past 20 seasons. Their record is one long good stretch. The past three seasons are the longest they've gone in twenty years without placing in the top 5 in the league.]
So. How many times did a good stretch follow a bad stretch by any length of time? Four. In only four cases out of thirty (13.3%) did any team go from being very bad to being very good over the course of twenty years. And that's if we're being generous about our definitions:
1992/93 - 1995/96
2002/03 - 2005/06
2001/02 - 2005/06
2007/08 - 2011/12*
1991/92 - 1996/97*
2007/08 - 2010/11
2003/04 - 2006/07
2008/09 - 2010/11
*Pittsburgh ranked 4, 8, 8, 3, 4 between 2007/08 and 2011/12, but they won the Stanley Cup in 2009, while placing 8th in the league in points. So we can reasonably call that a good year. San Jose placed 22, 23, 17, 19, 25, and 25 between 1991/92 and 1996/97. In the years they placed 17 and 19 (1993/94 and 1994/95) there were 26 teams in the league. 19th is just out of the range of a bad stretch but in the bottom 1/3 of the league at that time. 17th is in the middle of the league.
In only two of these instances did a bad stretch precede a good stretch by less than 7 years. In other words, two franchises in twenty years can make a case for being rewarded for very poor play with a dominant team built on those draft picks (Pittsburgh and Washington). Only one of those teams has won a Stanley Cup this way. Further, very poor play was followed more often by very poor play than it was by very good play. Six times, bad stretches were followed within 6 years by other bad stretches. That's three times as often as good following bad.
In fact, most frequently, mediocrity precedes winning and follows losing.
Three yrs before good stretch
Six yrs before good stretch
Three yrs after bad stretch
Six yrs after bad stretch
Three bad stretches are ongoing. One more ended less than three years ago. Two others ended less than six years ago. Those six stretches were not counted in these averages.
Both Pittsburgh (2009) and the Chicago Blackhawks (2010) won a Stanley Cup three years after the end of a bad stretch. But most of the time that is not the outcome of a bad stretch. Most bad stretches led to either mediocrity or to more bad years.
What if we change our definition of the reward for tanking/incompetence? What if the reward is winning the Cup? Could teams that play badly to get draft picks expect to win a Cup with those draft picks? Well, no. The average rank three years before a Cup year is 10.3. Six years before it's 11. The average Cup team in the last twenty years experienced only 1 bottom-5 year within six years before winning the Cup. Most Cup winners built up their teams over time, rather than all at once with top-5 draft picks.
In the last twenty seasons, only one team has managed to turn things around in a way that can be considered related to drafting very high. Put another way, in twenty years, the draft system has rewarded only one team for very bad play with tangible and long-lasting results.
One could argue that the Chicago Blackhawks won their 2010 Cup with high draft picks, but their actual performance in the two years since has been just as middling as the Calgary Flames' performance. They did not become dominant through high draft picks. They rolled the dice and won one year.
Which brings me to my point. General Management (or perhaps general management and coaching) appears to have a far greater impact on a team's fortunes than the draft. Very, very good teams stay good for a long time (Detroit, New Jersey Devils), while very, very bad teams teams tend to stay bad for a long time (Columbus Blue Jackets, Edmonton Oilers, Florida Panthers, New York Islanders) regardless of drafting.
Most teams operate somewhere in between those extremes, some having more good years than bad, others having more bad than good, but usually moving in and out of the middle of the pack. The average number of seasons that teams have spent out of the top or bottom 5 in the league was 12.
And it is much easier to stay bad than to stay good, which may be skewing our perceptions of how teams are playing. Only three teams have had no bottom-5 years at all in the past twenty seasons: Detroit, New Jersey and the Minnesota Wild. But the Wild have had no top-5 seasons, either. And they have seven companions in that: Calgary, Columbus, Edmonton, Florida, the Los Angeles Kings, the Islanders, and the Winnipeg Jets/Atlanta Thrashers. In addition, the Lightning are the only team to have had only one top-5 season.
In essence, rather than worrying so much about the draft rewarding teams for tanking or incompetence, fans ought to be worried about our general managers. #FailForNail campaigns can help you get through a season, but if you're having them more than once or twice, you ought to be concerned. And if your general manager looks at the lay of the land and decides that tanking is a viable strategy, he deserves to be fired. Because it doesn't work. And it took me all of four hours sitting at my laptop to see that.
There may very well be other incentives for a General Manager to resort to tanking. There is some evidence that getting a number one draft pick does have attendance benefits. But along with that come the risks inherent in a bottom-five finish, namely a general manager's job security. I can't say that tanking doesn't happen. I can say, however, that the draft system currently provides very weak incentives to intentionally play badly and rewards failure relatively poorly as well.