Shams: Spurs Trade for Reggie Bullock and Unprotected 2030 First Round Pick Swap

[Ariel]

he did than deleted the tweet as some picks were going to the Mavs etc., but those were the Spurs second rounders specifically included in the deal

Thanks, it's likely those are the picks involved then. Didn't catch it.

[Mr. Body]

Swaps don't have the value people are fantasizing they do. They're nice and can obviously hit every once in a while. But they're getting blown up because that's all teams have now. Like for example, them being unprotected is much more common than them being protected. Harping on "unprotected swaps" is like harping on unprotected seconds. Sure you can protect them, but the default is that they are free and clear. The Spurs made a shrewd move, especially if it's just those redundant seconds. Memphis made a meh move that will probably look stupid sooner than later.

I'm conservative with assets and would prefer a pick with protections rather than a swap. Then I have two picks rather than one + the relative chance that one pick moves up.

If I'm reading correctly, Wizards get to swap with Phoenix on both these picks (I believe) and then Memphis has the chance to swap with whatever Phoenix has at that point. Next year is highly unlikely to matter. The 2030 pick might be okay, but isn't going to be some genius pick unless both Washington and Phoenix manage to be terrible and pop in the high lottery.

I kind of see this as a waste of Memphis's second rounders. Phoenix, meanwhile, is very creative with already swapped picks. I can imagine how byzantine these swaps can get.

[scott]

In theory, PHX can trade a 2030 swap up to 29 times, and effectively lock themselves into the 30th pick! The NBA has entered its derivative trading era!

[Chinook]

Maybe - again, I think its savvy like my someone with money buying a few lotto tickets for fun. You don’t want to rely on them to make rent but if you are just having some fun it could pay off with little downside.

I know you are a bird in the hand type guy so you value the 2nds because they are for sure; but teams are having to give up things they probably rather would not out of pure need and I think theres value right now in the 2nds for swaps things. But that’s just me.

I think Memphis should keep their seconds. They've traded away a lot of them already. The Spurs have a ton, still. Like I think they're at the same number they were heading into this off-season. They can consolidate. Memphis is instead sacrificing flexibility for a tiny increase in impact. They absolutely have to be looking to make an all-in deal to justify the move. Like they probably got way too many seconds from Houston to sort of offset this. But they're still running a huge risk of just burning those picks if they can't get preemptive value out of those swaps.

[scott]

Yeah, a "2nd Generation" Swap, as I'll call these, are definitely far less valuable than the unprotected variety the Spurs got. I'd argue there is an exponential decrease in value in each Swap generation. The 2nd Gen is worth less than half of a truly unprotected swap, a 3rd gen would be worth less than a 3rd of the 1st gen, etc. The value* likely flattens out around the 15th gen.

*By value we are really talking Expected Value of the resulting pick for the team holding the swap, and by around the 15th pick you're EV would undoubtedly be somewhere in the mid to late 20s already, so you would never really make it to the ultimate 29th generation because there is no more value for anyone to extract. In practicality, there really isn't enough value even for the 3rd generation for a trade to commence.

[timvp]

HOLY look what MEM just did I wish we could have done this damnnnnnn

Eh, yeah like others have said, that's a tougher one to evaluate because the Grizzlies have to be better than both the Suns and Wizards for it to pay off. Even then, you'd need the extreme luck of both the Suns and Wizards being a lottery team in order for it to be a big-time win. In San Antonio's version, it just matters what the Mavs do.

I can't think of a way to fairly model it but logically it sounds like this Memphis trade has a lot lower possibility of becoming a lottery pick than the Spurs trade.

[DPG21920]

Eh, yeah like others have said, that's a tougher one to evaluate because the Grizzlies have to be better than both the Suns and Wizards for it to pay off. Even then, you'd need the extreme luck of both the Suns and Wizards being a lottery team in order for it to be a big-time win. In San Antonio's version, it just matters what the Mavs do.

I can't think of a way to fairly model it but logically it sounds like this Memphis trade has a lot lower possibility of becoming a lottery pick than the Spurs trade.

I guess I look at it like Spurs gave up 4 2nds and took on 10M for 1 (albeit better). MEM gave up 3 2nds and got 2, with the one in 2030 being pretty damn interesting with WaS being terrible and Suns having no picks to replace KD/Beal by then?

[scott]

Eh, yeah like others have said, that's a tougher one to evaluate because the Grizzlies have to be better than both the Suns and Wizards for it to pay off. Even then, you'd need the extreme luck of both the Suns and Wizards being a lottery team in order for it to be a big-time win. In San Antonio's version, it just matters what the Mavs do.

I can't think of a way to fairly model it but logically it sounds like this Memphis trade has a lot lower possibility of becoming a lottery pick than the Spurs trade.

We could build a model for the probably if we just assume that every team has an equal shot of being good or bad at random. So in any given year, a team has a 47% chance of being a lottery team. So for the Spurs, they need to be better than the Mavs, and then they have a 47% chance of getting a lotto pick.

For Memphis to be in the money with a lotto pick, they have to be better than two teams (PHX and WAS) and then they have to hope both teams are lotto teams, so 0.47^2 = 22% chance.

This is very rudimentary math, and of course teams are not lottery teams at random so it isn't perfect, but it does illustrate the exponential decline in the EV of the pick with later generations of swaps.

Like I said, this is the NBA's derivative trading era. s gonna eventually hit the fan like Countrywide Homes in the housing crisis for whatever team is left holding the bag

[Extra Stout]

Does Vegas allow betting on whether or not a protected traded pick conveys in a given year? Because that’s the frontier right there.

[spurs10]

In a swap does the team with the best record get to swap with a team with a worst record and better pick?

[scott]

In a swap does the team with the best record get to swap with a team with a worst record and better pick?

That's correct on how the swap works. And they only work in one direction, so when you trade a swap, it isn't a guaranteed swap no matter what - it only works for one party (who receives the swap) to get the better pick.

[scott]

Does Vegas allow betting on whether or not a protected traded pick conveys in a given year? Because that’s the frontier right there.

The team and Enron is salivating at the mouth.

[scott]

Just wait until teams start going towards SRP/FRP swaps - that's the ultimate. You get the right to swap your SRP with another team's FRP. The gold standard is house derivative assets.

[Extra Stout]

Just wait until teams start going towards SRP/FRP swaps - that's the ultimate. You get the right to swap your SRP with another team's FRP. The gold standard is house derivative assets.

How about allowing player swaps? Or swapping picks for players? Or combinations? Like you want the 5th pick and offer the 9th pick and your starting small forward? The possibilities are endless.

[Kindergarten Cop]

In a swap does the team with the best record get to swap with a team with a worst record and better pick?

It probably goes without saying, but just to clarify - the swap only takes place if the team who "owns" the swap has the better record. If said team finishes with a worse record, they obviously have no desire to give up their slot in the draft. The example with the Spurs gaining the right to swap with the Mavs in 2030 only happens if the Mavs have a worse record.

That's correct on how the swap works. And they only work in one direction, so when you trade a swap, it isn't a guaranteed swap no matter what - it only works for one party (who receives the swap) to get the better pick.

EDIT: What he said.

[scott]

How about allowing player swaps? Or swapping picks for players? Or combinations? Like you want the 5th pick and offer the 9th pick and your starting small forward? The possibilities are endless.

Contract swaps. We've gotten a contract swap with the Lakers. We then trade for Luka. We exercise our contract swap rights and now Luka has to play under Reaves's 4/56 and the Lakers are stuck paying Reaves the Supermax.

[Kurik]

Eventually it will become swap a pick in 2040 and we’ll lose tomorrow’s game against you.

[Ariel]

We could build a model for the probably if we just assume that every team has an equal shot of being good or bad at random. So in any given year, a team has a 47% chance of being a lottery team. So for the Spurs, they need to be better than the Mavs, and then they have a 47% chance of getting a lotto pick.

For Memphis to be in the money with a lotto pick, they have to be better than two teams (PHX and WAS) and then they have to hope both teams are lotto teams, so 0.47^2 = 22% chance.

This is very rudimentary math, and of course teams are not lottery teams at random so it isn't perfect, but it does illustrate the exponential decline in the EV of the pick with later generations of swaps.

Like I said, this is the NBA's derivative trading era. s gonna eventually hit the fan like Countrywide Homes in the housing crisis for whatever team is left holding the bag

Since it's such a long time period, one could assume equal chances at any given slot for all teams. You could also take any arbitrary distribution, if you're relatively confident in your estimation of where you'd be (with a given mean and variance). Assuming performance of all teams is independent, you can draw a joint probability distribution for each of the possible combinations, and then iterate though them all, calculating for every scenario what the gain is (you need to use a model for pick value, not all jumps are created equal, i.e 3 slots from 30 to 27 is a lot different from 3 slots from 4 to 1), and what its likelihood is, and if you ac ulate that product throughout the entire sample space you'll get the expected value of the swap. I did it for 2 teams, it's easy to generalize it to more teams, but the problem I found was that the models for valuation of slots are awful. The most accurate (not that I did an exhaustive search, I googled a few) was Pelton '17, the rest are horribly flat and do not account for the difference in value between top picks and the rest (for instance in Pelton's model #1 ~ #2 + #21, it's obvious #21 doesn't have that value and it's worse for the other models... I suppose I could go through all recent trades to draw a more proper model, but it's too time consuming for me right now. But those are the basics of such a model, I'll play a bit and if I like what I get I might post the results.

[scott]

Eventually it will become swap a pick in 2040 and we’ll lose tomorrow’s game against you.

YES NOW WE ARE TALKING. This is what happens when you let gambling companies be involved with sports

[timvp]

We could build a model for the probably if we just assume that every team has an equal shot of being good or bad at random. So in any given year, a team has a 47% chance of being a lottery team. So for the Spurs, they need to be better than the Mavs, and then they have a 47% chance of getting a lotto pick.

For Memphis to be in the money with a lotto pick, they have to be better than two teams (PHX and WAS) and then they have to hope both teams are lotto teams, so 0.47^2 = 22% chance.

This is very rudimentary math, and of course teams are not lottery teams at random so it isn't perfect, but it does illustrate the exponential decline in the EV of the pick with later generations of swaps.

Like I said, this is the NBA's derivative trading era. s gonna eventually hit the fan like Countrywide Homes in the housing crisis for whatever team is left holding the bag

My man. It was too tempting to not model it. After some quick and dirty modeling, assuming all teams will randomly get a pick between 1 and 30, the Grizzlies trade results in an average improvement of ~2 spots and the Spurs trade results in an average improvement of ~5 spots. The chance of the Grizzlies improving 10 or more spots is ~10% and the Spurs chances are ~25%.

So, yeah, similar trade but San Antonio's version is worth quite a bit more. Their improvement odds are more than twice as good -- plus the Grizzlies version can obviously never result in landing a No. 1 pick via swap. Factor that in and, yeah, there's a pretty big difference, tbh.

[scott]

Since it's such a long time period, one could assume equal chances at any given slot for all teams. You could also take any arbitrary distribution, if you're relatively confident in your estimation of where you'd be (with a given mean and variance). Assuming performance of all teams is independent, you can draw a joint probability distribution for each of the possible combinations, and then iterate though them all, calculating for every scenario what the gain is (you need to use a model for pick value, not all jumps are created equal, i.e 3 slots from 30 to 27 is a lot different from 3 slots from 4 to 1), and what its likelihood is, and if you ac ulate that product throughout the entire sample space you'll get the expected value of the swap. I did it for 2 teams, it's easy to generalize it to more teams, but the problem I found was that the models for valuation of slots are awful. The most accurate (not that I did an exhaustive search, I googled a few) was Pelton '17, the rest are horribly flat and do not account for the difference in value between top picks and the rest (for instance in Pelton's model #1 ~ #2 + #21, it's obvious #21 doesn't have that value and it's worse for the other models... I suppose I could go through all recent trades to draw a more proper model, but it's too time consuming for me right now. But those are the basics of such a model, I'll play a bit and if I like what I get I might post the results.

I've always thought this was the best attempt at quantifying the value of picks: https://thedatajocks.com/nba-draft-pick-values/

[Ariel]

I've always thought this was the best attempt at quantifying the value of picks: https://thedatajocks.com/nba-draft-pick-values/

I like it much better. I tested an unprotected swap (not simulation, theoretical value) between 2 teams and assuming equal odds at each slot for both, I got that:
1) Using Ashbrock model for valuation (yours), an unprotected swap is roughly equivalent of a #19 pick
2) Using Pelton model for valuation, an unprotected swap is roughly equivalent of a #30 pick (this model also states that #1 = #2 + #21 so take this with a grain of salt)
If anyone wants the code (it's extremely simple) I can share. It's a bit late so I hope I didn't mess up anything

[timvp]

Since it's such a long time period, one could assume equal chances at any given slot for all teams. You could also take any arbitrary distribution, if you're relatively confident in your estimation of where you'd be (with a given mean and variance). Assuming performance of all teams is independent, you can draw a joint probability distribution for each of the possible combinations, and then iterate though them all, calculating for every scenario what the gain is (you need to use a model for pick value, not all jumps are created equal, i.e 3 slots from 30 to 27 is a lot different from 3 slots from 4 to 1), and what its likelihood is, and if you ac ulate that product throughout the entire sample space you'll get the expected value of the swap. I did it for 2 teams, it's easy to generalize it to more teams, but the problem I found was that the models for valuation of slots are awful. The most accurate (not that I did an exhaustive search, I googled a few) was Pelton '17, the rest are horribly flat and do not account for the difference in value between top picks and the rest (for instance in Pelton's model #1 ~ #2 + #21, it's obvious #21 doesn't have that value and it's worse for the other models... I suppose I could go through all recent trades to draw a more proper model, but it's too time consuming for me right now. But those are the basics of such a model, I'll play a bit and if I like what I get I might post the results.

Good stuff.

Thinking about it more and factoring in that not all picks are created equal, the Spurs version of the deal looks a lot better than the Grizzlies version.

The Spurs end up with the No. 1 pick from the Mavs 3.3% of the time. The Grizzlies can't get a No.1 pick from their trade.
The Spurs end up with the No.1 or No. 2 pick from the Mavs ~6% of the time. The Grizzlies end up with a No. 1 or No. 2 pick from their trade about 0.41% of the time.
The Spurs end up with a No. 1, No. 2 or No. 3 pick from the Mavs ~9% of the time. The Grizzlies end up with a No. 1, No. 2 or No. 3 pick from their trade about 0.9% of the time.

So, yeah, the trades sound similar but the Spurs have a ~14 times better chance of ending up with a top 2 pick from the Mavs and a ~10 times better chance of ending up with a top 3 pick from the Mavs compared to the Grizzlies odds. Considering that the value of most drafts is at the top, it's safe to say that the Spurs trade is like, what, 7 or 8 times better than the Grizzlies trade? Maybe 10 times better depending on how much extra value you give getting the top pick?

Basically, it looks like the Grizzlies rushed to do a copycat version of the Spurs trade but didn't take the time to run the numbers.

[heyheymymy]

cool forecasting analysis, thanks

[ismael-robert]

Great googly moogly! Thankful I just enjoy watching the games