The NHL and NBA annual entry drafts have become strange affairs. The lottery system (and their ever-changing weights) have at times encouraged fans of meddling teams to urge their favorite team to underperform in order to have a higher probability of selecting a top prospect. This is known as tanking or, more euphemistically, re-building or re-tooling. Many rational fans suggest that if you are not going to win a championship, you might as well maximize the chance of adding top young (cost-controlled) talent. Under the current system, they aren’t wrong much of the time.
There are no easy solutions to such a problem because there are two opposing forces at play:
1)The goal of the entry draft is to distribute new talent fairly throughout the league. Ideally, the worst teams should have an opportunity to draft the best talent, giving them an opportunity to compete in the future.
2)The goal of the league is to maintain a competitive product throughout the season. In a world where the incentive to win is diminished, the league product and brand suffer.
A lottery makes some sense. Teams can lose on purpose, but that still doesn’t guarantee the top pick. Would ‘rebuilding’ teams strip down their roster and be satisfied with a top 5 pick? Probably. Would the same team completely throw a few games to increase the probability of drafting 1st overall by 5%? Probably not. Draft lotteries use randomness to uphold a general competitive balance within the league.
Tanking it to the extreme?
However, very few teams are happy with the current system. This is a function of dumb luck and the perceived abuse of the entry draft system. Research suggests the value (the average quality of player historically drafted in that position) of a draft pick decays non-linearly. That is, the difference in value between the 1st overall and 2nd overall is greater than the difference between the 2nd overall and 3rd overall and so on. If you can’t compete, you are better off trying to maximize draft pick value by taking advantage of this non-linear curve – get the highest pick possible.
With very few happy with the current system, alternatives have been suggested and gained some traction (most recently the Gold Plan). However, year to year lotteries with 30 teams will never appear completely random to the human mind, so there inevitably will be annual disappointment with the system from all but one or two fanbases.
The Entry Draft Auction Proposal
The parameters below require more research, but more importantly, must be sold to the owners and teams. An expanded rationalization and methodology can be found later.
Each team receives a set amount of draft currency based on their finish during the regular season
The worst team would receive 1,000 base draft units (to be coined Bettmans in the NHL). Rank ordered from worst to best each subsequent team would receive 10 fewer units, meaning the champion would receive 710 draft units.
Each team would have their base draft units adjusted by the z-score of their offensive production multiplied by 10. A team can receive more draft units than the team below them by out-scoring them by a significant amount (approximately 23 goals in the NHL).
The team’s maximum bid is set to the number of their base draft unit plus offensive production adjustment draft units. This prevents bottom teams from selling current assets in order to secure enough draft units to guarantee to win the bid for the top pick (this would be a terrible strategy unless there was a generational talent available, but still).
Draft units by year could be traded in absolute, share of total, or conditional amounts.
On draft day each pick or draft slot is auctioned off in real-time at the draft. The number of draft slots available remains unchanged from the current system.
Bids of whole units are blindly and simultaneously submitted. Ties would go to the team with the fewest number of picks to that point, then highest number of picks since prior team pick, else re-auction between tied teams. Losing teams would lose no draft units. The winning team would lose their bid amount, or alternatively the value of the 2nd highest bid.
Rationalizations & Methodology
A further explanation of some of the main ideas behind the Draft Auction.
Base Draft Unit Allocation
The initial allocation of base Draft Units requires research and agreement from all parties. I think there are different ways to do this but here is my framework.
In the NHL the team with the top draft position can be expected to recoup about twice as much talent as the championship team. The shape of the curve also suggests the value by position decays again in a non-linear way. Should the difference in expected value received by the worst team and 2nd worst team be greater than the difference received by the 2nd worst and 3rd worst team? Probably not in a fair system. Also consider that these positions were likely arranged by a lottery.
I would argue that a more equitable system would decay team-level draft value linearly. This can only be accomplished by assigning the granular draft units proposed above. The graphic below re-enforces how this could be accomplished by distributing draft units.
An auction system and agreed upon distribution of draft units would also allow the league to close the gap between the expected value received between the top and bottom teams. The worst team receiving twice the expected draft value seems excessive in the age of salary-cap assisted parity. Expanded research might answer this question in a quantitative manner, but realistically the robustness of the research would take a backseat to a buy-in among the 30 teams. The Auction Entry Draft proposal sells the idea of a more equitable and competitive league (and nobody envisions themselves being the worst team in the league) so it seems like there would be support for closing the gap.
I would suggest the championship team should receive 71% of the draft value of the worst team (see chart above). This is the result of easy math – each team receives 10 fewer base draft units than the team immediately below them in the standings. The formula could be expanded to account for non-playoff team ties, distributing points throughout the league from the worst to best teams in a linear fashion.
Bonus Draft Units for goals scored
Yes, it is kind of video game-y, but there are 2 reasons I think it would be worthwhile:
1)Add some noise to the system. If a generation player came along that you would trade your entire draft for, the last place team couldn’t sit on their standing position and out-bid everyone, other bottom feeders could outbid them if they out-scored them by the appropriate margins. Obviously, everyone is trying to score the maximum amount of goals anyway, this just keeps teams honest.
2) Incentivization of higher scoring strategies – this is generally good for excitement.
Between 2007 and 2016 (excluding the lockout-shortened 2013-14 season) team’s scored an average of 222 goals per season, with a standard deviation of 23. Full equation:
Below are the distributions of Goals For z-scores in the last 8 season. By definition, 68% of teams would not have their draft units adjusted by more than 10.
Alternatively, this calculation could use goal differential. Or there could be no adjustment, again the concern would be a team could theoretically tank and guarantee the right to draft a generational talent with their entire draft stock. Adding an adjustment would prevent this strategy.
Can’t mess with Trade Deadline Day. Teams can get even more creative since there is no draft picks constrained by round and standing. Want to trade 100 draft units? Great. Trade for 10% of the other team’s base draft units? Cool. 500 draft units if they make the Cup final, 200 otherwise? Sign right here.
Limited Number of Auction Slots
This is more of a Players Association issue. A cap on the number of auctions keeps the number of drafted players the same or fewer. A case where draft units are not properly rationed and there were no teams left to bid on the last few picks would be a generally good thing.
Real-Time Auction System
This is where I give pause. Entry drafts are high profile events, with lots on the line. The technology component would be critical, any failure would be embarrassing and would require the right safeguards. Every team would have 30 seconds to submit a bid, the winner or a re-auction would be announced immediately, and a mandatory 30 more seconds would pass in order for any team with a technical objection to raise it to officials, then the winner would be on the clock to make their pick.
Entry drafts have traditionally been monkeys throwing darts at a dartboard, right? So why add another layer of complexity?
Well, economists love auctions because of their ability to imply value, particularly hard-to-calculate values (like the right to draft an unproven, underdeveloped teenager). An auction-based system would be a bonanza of implied information all while being highly entertaining.
It would also further the encourage the operational analysis that has recently grown in sport. Drafts would be fueled by both computer models and high-drama gambles. Data at the draft slot-team level could be made available to teams and public allowing for a unique look into the question – how do teams value draft picks? The trend is clear – advanced analytical methods are becoming the norm in sport, and this system would only accelerate that healthy trend.
Most teams would struggle to neatly quantify the value of draft auction (factors could include, but not limited to, talent currently on the board, total amount of draft units in circulation, current team draft units, historical valuation of draft slot), but it would be a beautiful mess of varying strategies with plenty of unforeseen events. Poorly managed teams would struggle with this configuration but the incentive is clear under an auction system: organizations must commit to competing annually, provide an exciting product, and leverage analytical methods.
The Entry Draft Auction would:
Remove the incentive to tank, distributing talent in a more equitable, linear way
Incentivize offensive strategies, increasing quality of product
Create a unique and highly entertaining experience, producing highly informative data
Create more granular and flexible trade blocks, helping facilitate trades and optimal talent distribution around the league
Thanks for visiting the CrowdScout blog –Game Theory!
The CrowdScout platform was designed to automatically and elegantly aggregate the opinions of awesomefanalysts and create unique content – dynamic player rankings that can:
·aid the decision making of managers (fantasy or professional)
·settle arguments that happen over cold ones (or not)
·provide benchmarks for more advanced analysis, i.e. when determining what players are over/undervalued
·identify scouts with the ability to be ahead of the curve on judging talent
The inaugural beta season was a great learning experience, and I have some exciting plans for season 2, but clearly the website hasn’t hit the critical mass to provide dynamic and self-sustaining content. To supplement the CrowdScout system, I’ll be throwing out some of my own thoughts in my Game Theory blog.
What is Game Theory (or what will it be)?
·Hopefully delivers both qualitative and quantitative insights to sports (predominantly hockey) – the original idea behind the CrowdScout platform
·Part thought experiment, part analysis. Some logic and some numbers
·Whatever seems interesting and easy to write to me. If it is boring to write, I can’t imagine how bad it would to be to read
·Ideas a little different than the standard – meant to be critiqued. Ideas are stronger with more diverse input – one of the main principles behind CrowdScout
·Potentially more advanced analysis, possibly combining my own proprietary CrowdScout data with public data
I’ve been lucky enough to live lives of a colligate hockey goaltender, an antitrust economist, and a data scientist. I plan on relying on the ensemble of my experiences rather than one – there are more interesting economists and statisticians discussing sports worthy of your time (I believe market forces have spoken on my goaltending abilities as well – unless the NHL was really serious about increasing goals). After finishing my college hockey career, I took some time away from being completely immersed in hockey while the hockey analytics community matured. When I decided I wanted to contribute, I thought it would be best to create something different – a platform that was able to combine analytic and traditional information in a meaningful way. I hope to do the same in the Game Theory blog.
 Full disclosure: I played Division III NESCAC hockey (only because there was no Division IV, as my coach liked to remind me)
 I also fell ass-backwards into doing anti-trust economic consulting and advising the most recent NHL lockout – a bitter-sweet, but very exciting experience