2019 NBA Draft Odds by Win Projections
Everyone loves the NBA draft, especially when your team possibly owns four picks. 2018-19 season projections have been coming out so I figured since I don't have the time/desire/math skills to make my own projection model, I would use the data from someone else's and project out 2019 draft position odds instead.
Jacob Goldstein was nice enough to provide the world with the projected full season win totals that 10,000 simulations of his model produced, so that's the source data for everything that follows. I also use the new and old lottery tables and Kevin Pelton's Pick Value Card to show the relative worth of different teams' draft stock in 2019.
Full NBA Odds
Displaying the full odds table is difficult just because it's a lot of numbers. You can scroll over in the embedded table to see specific results. The "PeltonValue" is the weighted value of a team's 2019 selection (or the protected/possible parts of a selection) with 4,000 being the value of the #1 pick in an average draft.
Jacob's model has the Knick, Hawks, and Kings as all equally bad teams. However, the Kings pick, which is owed to either the Celtics or 76ers, comes out as being slightly more valuable than the others because they're in the West and so more of their higher end outcomes still land them in the lottery.
The picks originating in the East make up 52% of the total Pelton Value of the selections. They account for 49% of Jacob's win projections. In the coming season, Jacob projects the 8-seed diving line to be about the same number of wins in the East and West. If one conference had a significantly higher 8-seed projection it could skew things so that a relatively small amount of additional pick value would flow to the strong conference. At least based on Jacob's preseason projections, that's not an issue for the coming year.
The 2019 lottery sees the implementation of some anti-tanking reforms. The system is changing to "flatten" the odds and to draw the top four picks instead of the top three.
We can see how this has some impact on the projections of each team. As expected from an anti-tanking measure, the worst teams take the biggest hit in odds and value.
The team that actually suffers the most from the changes is Philadelphia because they would get the Kings' pick if it lands at #1 but not 2-4 (except in the extremely unlikely scenario where their own pick is better than Sacramento's). Because they're dealing only with top-1 protection, they see the largest value decrease. Of course, they knew the changing odds when they made the trade so this isn't really a case of anyone being "hurt" by it.
Atlanta and the Knicks see the next most dramatic drops in value. While those two teams have the same average wins in the projections, Atlanta's win projections are spread out more and so the system thinks that the Knicks' lottery seed is somewhat better.
Memphis and Dallas see the most benefit of the changes because they have pick protections at work and the flatter odds could push them into those ranges. Dallas made their trade after the new odds were known but Memphis's trade predates the updates. In both cases the pick obligations would roll over if the protection hits so the total actual values aren't what's listed for just 2019.
The teams that truly get the most benefit are the ones projected for the mid-to-late lottery. The expected value changes aren't that big, but of course if one of those teams actually does hit the jackpot it could be franchise changing.
The Celtics have the possibility of getting four picks in the 2019 draft. They currently own their own plus (barring another trade) will get either Sacramento or Philadelphia's. They get the better of the two picks, unless one of them is #1 in which case they get the lesser.
They will also get Memphis's pick if it falls outside the top eight and the Clippers' selection if it's not in the lottery/top-14.
Declaration of Independence from the Kings Pick
With the protections as they are, Boston will likely get the Kings pick. The lottery could be very dramatic though, as the downside risk of that pick landing at #1 and going to the Celtics' most direct long-term rivals is huge. Adding insult to injury, Boston would likely have a pick in the 20's coming their way instead. Based on Jacob's current projections, there's a 12% chance of that happening, which is too much for my liking.
The Celtics project to having the #5 pick in nearly 20% of scenarios, in large part because of the new top-4 lottery system.
Grizzlies Pick (Jeff Green Trade 1)
Jacob's model projects Memphis to be just good enough to likely send their pick to Boston. It would be no surprise if they don't, especially considering the new lottery odds and their injury exposure, but at the moment that's what the simulations say.
An argument can be made that a team in contention should want the pick to convey as soon as possible so that they can get a helpful player in and developing. However, with the state of the Celtics' depth and cap sheet, plus the high upside of diminishing pick protections, they would probably rather have the pick roll over to 2020 or 2021.
You can see the impact of the new lottery odds in Memphis's projected spot in the late lottery. They're more likely to have a top-4 pick than one in the 5-8 range if they do keep it.
While the 8-seed diving line is projected to be close between the East and West, the West line is still modeling out at a higher number which is why the Grizzlies odds of sending specifically the 15th or 16th pick is lower than the 17-19 range. That's where the West 7 and 8 seeds are most likely to land in the draft.
Clippers Pick (Jeff Green Trade 2)
Boston is much less likely to get a pick from the Clippers because the obligation is top-14 protected. The Clippers and Grizzlies are projected to be similar level teams, though, so the shape of the draft odds are the same. It's just that the line where the pick conveys moves over.
The total Pelton Value of the pre-lottery / pre-protection Memphis and LA picks are nearly identical but that protection level dramatically shifts the value for Boston.
The Clippers pick protection does not weaken next year and if it doesn't convey by 2020 it turns into a second round pick so, unlike with the Grizzlies pick, Boston hopes that this one does transfer in 2019.
Boston's own pick is likely to sit in the mid-to-high 20's. Jacob's model favors Toronto as the 1-seed and has Philadelphia not that far off the Celtics. The historic value difference between pick 25-30 is tiny, though. It matters a lot more for the playoffs than the draft.
With the Celtics having at least one other pick guaranteed, I think their own 2019 pick is more likely to be traded than in past seasons. That could happen in-season once they know more about how the Kings, Grizzlies, and Clippers are shaping up, or could be used at the draft to move up if one of Memphis or LA do send a mid-round pick. No specific trade is every particularly likely, but Boston could end up with more picks than open spots and, unlike 2nd rounders, these would all carry guaranteed scale offers.
Multiple Pick Chances
As of now, Boston is guaranteed two picks and could have as many as four. Based on Jacob's projections, the most likely scenario is having three picks.
The ideal scenario of getting the Clippers' pick but having Memphis roll over the 2020 is the least likely outcome of the four possibilities.
These numbers are all based on Jacob's model which I can't validate but his expertise and reputation are better than mine, so that's good. The lottery odds are known. I have my own pick value card but use Pelton's (much) more well known version and the two are similar, anyway.
Jacob's model is based on 10,000 season simulations and then I placed teams in the league standings (to set if they were in/out of the lottery) by using those win totals to run another 20,000 simulations off random number assignments. If I submitted this for a PhD I'd be in trouble, but it works for this purpose well enough.
I did take some shortcuts in this analysis, though.
First, I broke all standings ties with a random number drawing. In reality, if two teams have the same record and are both in the lottery the lottery combinations are shared for the tied spots. For example, two teams tied for the 6/7 spots in the lottery would each get half of the total combinations assigned to the two. Building multi-team tie logic would have been too much work, so I just force break every tie. Over the total data set it makes only a very small difference.
Second, I did some rounding that resulted in only 99.7% of scenarios being covered. I force-filled that 0.3% gap by giving every result a tiny bump of that amount. The difference is functionally un-noticeable.
As a check to make sure I don't have any obvious errors, bad formulas, typos, etc. I did a sanity check. The average projected wins and the Pelton Value of the projected picks should be in roughly inverse proportion. They are, and so I think the numbers are generally good.
Thanks to Jacob Goldstein for the wins projections.