Value of a Draft Pick and Changing Value in the NBA draft

Quick note the Free Agent Guide is updated

I originally had a long post on the Worst Team since the merger but events and a computer crash have forced me to delay that for this weekend. However, I have a goal. 365 days, 365 posts. To that end I will move up a post on answering some followups from my draft posts ( here and here)

Reader BPS caught an error in one of my posts and as a result won a no-prize. What that entails is that he gets to call his post. His request?

The thought I had when looking at this chart was to try and figure out the monetary value of a draft pick. Since the pay scale is fixed, and we have a rough idea of the expected production of each pick, and a value of a win per year…

This is a perfect question  and one that naturally comes up when looking at NBA draft data. What is what is the value of a draft pick. What’s the most advantageous position for my team? To answer this I built the following chart:

You're damn right this was a lot of work.

The table uses the compiled draft data, a value per win based on the 09-10 payroll of $1.7 million per win and the rookie salary scale for 09-10. The results are very surprising. If I rank the top 10 picks in order:

  1. Pick 1
  2. Pick 3
  3. Pick 5
  4. Pick 9
  5. Pick 4
  6. Pick 2
  7. Pick 7
  8. Pick 11
  9. Pick 10
  10. Pick 24

Pick 1 is still the best value but Pick 2 is all the way down at number 6. This happens because not only is incoming talent evaluation an inexact science (even more so than existing talent) but salaries matter. A 9,10,11 or 24th pick is much less costly than a 2nd pick.

One other interesting point about the draft is the presence of impact players.  What are the chances of drafting a star or a superstar and has that changed over time? Was the draft better in the past? Given the data set we can definitely chart this up. Some ground rules first. A .100 WP48 player who plays 2000 minutes will generate about 4 wins (4.167). We will look at each draft and divide it up into 5 groups of interest:

  • Produced <= 0 Wins a year (The Losers)
  • Produced > 4 Wins a year (Better than average)
  • Produced > 8 Wins a year (Star)
  • Produced > 12 Wins a year (Superstar)
  • Produced > 16 Wins a year (All Time Great)

Using these parameters the draft classes look like this:

1984 was a watershed draft class for the NBA (Barkley,Jordan,Hakeem and Stockton that almost enough for an all time team). But recent classes have not been bad (1999,2004 and 2005). But rather than individual years we want to see the draft over time. If we take this a step further and look at percentages and do a sliding average at a period of 5 years:

And the accompanying chart:

What we can see paints an interesting picture. Of note:

  • The probability of getting no value or negative value from a draft pick fluctuates around 30% , peaking at close to 40% in the period of 1987 to 1991 which interestingly coincides with the well know drug problems in the league.
  • Value peaks in the 82 to 86 period and tails off to a low in the mid 90’s but it’s picking back up in the most recent samples.
  • Impact players (Superstars & All time greats) are coming into the league at an increasing rate. This would help explain the fact that the quality of basketball seems to be at it’s highest levels in recent years (see here)

So what conclusions can we reach from this analysis? The draft is a high stakes lottery but it’s a rigged game for the owners. Salaries are fixed at a discount  and the risk of utter failure is relatively low (30%). Given the rising tide of talent that the data seems to point to, a small market gm hoping to stay competitive can use picks rather than free agents as the way to keep your team successful both on the court and in the bottom line (See San Antonio and Oklahoma City).

Written by:

16 Comments

  1. Chicago Tim
    7/24/2010
    Reply

    Yes, well, for the alternative conclusion that the way to win championships is to pursue free agents and forget about the draft, see Miami. See also L.A. when they pursued Shaq. But yes, the best of both worlds is to draft well and attract top free agents, who are also underpaid due to a cap. If you can do that, then the middle-tier players will accept discounts to play with a quality team, as we have seen this year with Miami and, to a lesser extent, with L.A.

    I see you added Barnes to the free agent update and that L.A. got a huge steal. In any other year I would call that unfair, but this year I’m just hoping the Lakers can give Miami a run for their money.

    I know we don’t have exact figures for the Bulls’ deal with Kurt Thomas, but I’ve heard one year for just above the veteran’s minimum of $1.35 million. I think when you add that to the mix the Bulls will leapfrog the Lakers, and I think Golden State as well, in the “winners” list.

    I’m also hoping the rumors are true that the Bulls have a team option for a second year on Thomas’s contract. Thomas is such a great pickup, not only because he can still play quite well, but also because he has a great attitude and can teach the Bulls’ other bigs. Plus, what a great price!

    • 7/24/2010
      Reply

      Tim,
      I should have added “for the small market gm”. It was late and I was sleepy. LA is actually a team that has done really well from the draft over time.

      Agreed on Thomas being a good pickup

      • Chicago Tim
        7/24/2010
        Reply

        Well, Miami is a small market, although it has attractions to NBA players that most small markets do not.

        Yes, LA seems to use both approaches, draft well and pursue free agents. And originally Wade came to Miami through the draft. I don’t think a GM should ignore either approach.

        But the overwhelming lesson of this season, I think, is that success breeds success, and failure breeds failure. A successful team will get discounts from free agents and will retain the players it drafts. An unsuccessful team will get stuck with overpriced players and will lose any great players it is lucky enough to draft.

        It’s very hard to turn around an image of failure. Perhaps the best way to do it is to pay top dollar for a competent GM and a competent coach, and then show a willingness to pay a reasonable price for great players. That doesn’t mean spend foolishly, but it does mean spending money on top talent in management and on the team.

        I would like to see the value of second round picks. This could help evaluate trades for lottery-protected draft picks or second-round draft picks that are often parts of trades.

        • 7/24/2010
          Reply

          Excellence breeds excellence. This is why when I’m brought in to flip a plant one of the first things I look for is cleanliness and organization (the japanese 5-s). The best mfg. plants are so clean and organized you could perform a medical procedure on the floor. It’s a pleasure to work at such a place.

          For the NBA the same rules applies. If you had a choice would you work for the Knicks or Clippers or for the Lakers or Celtics. The Hawks or Raptors or the Spurs, Suns and Mavs. I know how I would choose.

    • 7/24/2010
      Reply

      Guy,
      Thanks. Scanned it. It looks like a good piece for sunday reading. I like WS but I feel it tends to overspread wins around. This reduces it’s value when players change teams. This i think explains some of the discrapancies between the numbers there and here. I will take an indepth look though.

  2. 7/24/2010
    Reply

    Yet another great post Arturo, your blog is really beginning to rival WoWJ as my favorite basketball blog.

    Any you could post the first table as a google doc so we can play around with it?

    • 7/24/2010
      Reply

      Shawn,
      Thanks for the praise. I will see what I can do about that gimme a few hours.

  3. Dan
    7/24/2010
    Reply

    It looks like there’s a fair amount of random variation in the Wins Produced numbers – there’s no reason why the 15th and 19th picks should produce the fewest wins out of the top 27 picks, or why the 24th pick should produce more wins than any other pick outside the top 11. If you fit a curve to Wins Produced to smooth out the variation, then the value to the team (after subtracting the cost of the salary) will decrease monotonically, meaning that it’s always better to have an earlier pick – the added value of a better player is worth more than the additional cost of a higher salary. The 9th pick isn’t really worth more than the 2nd pick, it’s just that teams have had good luck with the 9th pick and bad luck with the 2nd pick.

    • 7/25/2010
      Reply

      Dan,
      You raise a good point. I really should look at value available or best value as opposed to what actually happened. So really the point should be that typically there’s value available in the draft late in the day.

      • Dan
        7/25/2010
        Reply

        You don’t need to change that much – I just think you should fit a curve to the data, since 30 drafts is too small a sample size to get rid of the noise in the data for the value of each pick. You could think of it this way: if you were predicting the value of next year’s draft picks, you wouldn’t predict the 9th pick to produce more wins than the 2nd pick. Instead, your predictions should follow a smooth decreasing curve, with each pick expected to produce fewer wins than the previous pick. So look at the data and try to fit that curve.

        Or, you could think of it this way: if you looked at the data from one draft, the values for the picks would bounce all over the place – you’d probably get some sense that earlier picks tended to perform better, but there would be a ton of noise. If you had 10000 drafts worth of data, you would expect the value per pick to be a smooth curve, monotonically decreasing, since the noise would be pretty much all washed out by the law of large numbers, leaving behind something very close to the true value of each pick. With 30 drafts of data the results are somewhere in between – they bounce around a fair amount, but still give a good sense of what that smooth curve would look like. I’m just suggesting that you should fit that curve to the data to better estimate the true value of each pick, rather than letting the random variability in a sample of size n = 30 get included in your estimate of the value of each pick.

        I tried out a few curves, using some simple transformations, and found that Wins Produced per pick = 31 – 8.1 x ln(pick number) fits well (R squared = .87, and no noticeable pattern to the residuals).

        • 7/25/2010
          Reply

          Dan,
          There are problems with the assumption that wins by pick are going to be monotonically decreasing.The draft is not a rational market. GMs demostrably do not pick the best available talent. The data in fact points to the variation being common cause and not special cause. Without a major shift in the talent evaluation model, I would not expect the data to shift irregardless of the sample size.

  4. todd2
    7/25/2010
    Reply

    Arturo qualified his closing comments by stating “competitive,” not winning championships. I’m looking forward to his post covering worst (and best, hopefully) post-merger teams. IMHO, Miami, New York and Boston’s feast or famine approaches do a disservice to fans who have to endure some lean years for flashes of brilliance. And let’s not give Heat and Celtic management too much credit; the player themselves played a significant part in their transactions.

  5. 7/25/2010
    Reply

    todd2,
    I have working theory on how the requirements of winning championships are no the same as those of winning in the regular season. That’s why I qualified the statement. This is something i’ll get to at some point.

Leave a Reply

Your email address will not be published. Required fields are marked *