Why don't we start there just to kind of get warmed up? Because there is some similarities with how these things behave with regard to portfolio insurance. So give us kind of the high level of when you think about the impact of products on prices involved, how do you incorporate things like the experience of the 87 crash?

So I did a very deep study on Paul Tudor Jones' work around portfolio insurance in 1987. And part of what's so interesting about that is that there are two kind of stories, right? One is the PBS special Trader, which if you ever get to watch is a fantastic snapshot in time, but it basically goes through this idea that he is modeling on the basis of the similarity between the 1929 stock market and the 1987 S&P 500 and how this is going to predict a crash. That is kind of the headline story behind some of it. But the much bigger story, and this is actually contained in his writings to his investors, if you go back and review his letters to his limited partners in the 1987 time period, what he was really talking about was the reliance that the market suddenly had developed around these products and how these products were consuming an extraordinary amount of the liquidity of the underlying at any point in time. And so effectively, he was saying portfolio insurance is not dissimilar to the subsidized flood insurance along the Mississippi or any other river along the coast of Florida that encourage people to make investments that they otherwise would not make. So you get into this process because you are convinced that there is a way to protect yourself or get out that allows you to take additional risk. And that process can actually create exactly what you're describing, the endogenous feedback loop that then amps up the risk and the exposure within the market. What Paul identified was very similar to some of the stuff we'll talk about going forward, which was those products themselves had in turn become so large that if you actually started a process of reversal, you tried to actually exit or execute the insurance policy, that it itself would create a feedback loop that would then force the feedback lower and lower and lower. Specifically in portfolio insurance, the underlying strategy was effectively, we are going to do the equivalent of delta hedge your portfolio by selling futures against it. As the market falls further, you have to sell more and more futures in order to reflect that the insurance is becoming a bigger and bigger portion of your portfolio. And there comes a point at which you try to execute the next sale and it's so large, it overwhelms the market. And one of the key untold stories of the crash of 1987 is one that came out when I was speaking to Mark Rubenstein in early 2009. I was reviewing some of the trades that I was looking at that point. I happened to have developed a relationship with Mark Rubenstein, who was one of the partners at Leland O'Brien Rubenstein, the largest of the portfolio insurance firms. Mark was a finance professor whose ideas around Delta hedging actually underpinned what's called the Cox Rubenstein model of options, the binomial pricing trees. And what Mark told me was that when it got to that point, the next sale was going to be so large, his trader actually came to him and said, Mark, if we execute this trade, we will send the market to zero. And Mark took the fiduciary responsibility and said, no, right? We're not going to do the trade. We will take the loss rather than send the market to zero. I think that's a really critical insight on two fronts, right? One is we were still existing in a world there where there was a human check. Do we do this trade? We have to execute this trade. Now it's largely embedded in algorithms. And so one of the things that you always have to think about is what would have happened if that human intervention hadn't existed? So that's one of the things that hits me every time I think about these components. I

also knew Professor Rubenstein. He actually was on the board of a company I had founded in 2000. We didn't get super far, but I got to know him well. And I just always found it quite fascinating that the Leland O'Brien Rubenstein part of this and just the take up of corporate America on this product was so substantial that, you know, in some ways, if you just looked at the sponsorship of the product and the Delta hedging requirements, again, you were going to have to get subcatalyst. I think the VIX got to 150 on the crash of 87, but it was 40 plus the week before. This thing didn't truly come out of nowhere. Obviously, it accelerated, but it's just a fascinating and in a lot of ways, the first example, at least in the stock market, of the market just overwhelming itself. And of course, there are many other examples. After that, you and I have shared an interest in a study and played some part of analyzing the XIV meltdown in 2018. And then around 2017, you could have done some pretty similar analysis and realized that, boy, these guys are going to exhaust the liquidity available in VIX futures pretty quickly with a shock that in some ways was difficult to imagine because 2017 was so low vol, but it really wasn't all that big a shock. No. And even

more recently, we actually just saw the unwind of the dispersion trade around August 5th, in which a lot of people have misassociated August 5th with the Japanese yen. I would argue that was just the dispersion trade where people are selling volatility on the index and buying volatility on single stocks. The easy articulation of it is the market cannot go up unless NVIDIA goes up. Therefore, I'm better off buying increased exposure to call options on those individual names and selling that exposure to the S&P. When the actual unwind occurred, the implied volatility to unwind that, I have to buy back my exposure to the S&P while selling it on single stocks. And we saw those relationships flip. And I think you and I talked about it at the time. You saw the number, quote, unquote, implied correlation briefly went to 240%, right? Now, anyone listening to this type of broadcast would know that correlations actually can't go beyond 100%, but that implied pricing is telling you how out of whack things got.

Yeah. What a dislocation that was. Let's talk about that briefly. It kind of came and went. I was just looking at one month implied vol in the S&P. Last three years, it's in the kind of first percentile. Now, look, it's early December. You've got Christmas. You've got New Year's. So it's a holiday period. It definitely gets quiet towards the end of the year. So not a huge surprise, but quite a difference from that August 5th disruption. What's the big picture? What should we take away from that asymmetry in terms of the VIX clearing at a price and then suddenly it gapping out to such an extent?

Well, I think that there's a couple of things that I would take away from it. One is March 2020 is recent enough in history that I would argue that things like the dispersion trade and the short volatility complex, as large as they have gotten, are very different than the types of risks that were being taken going into either 2018 or certainly March 2020, which saw a number of firms just taken out of that kind of uncapped variance exposure. The second thing that is important to remember about volatility is that volatility actually is a disappearing event. So a VIX contract is a contract for 30 days of implied volatility on the S&P 500. As you introduce things like one-day options, et cetera, you're seeing more and more concentration of actual hedging risks tied to the individual event. So if I know Jerome Powell is speaking on Tuesday, I will hedge Jerome Powell on Tuesday through a one-day VIX option that can be constructed through the existence of the S&P single-day option components. That means there's much less pressure on that VIX complex. There's much less need for it, which was why it was kind of concentrated in this dispersion trade, which had taken on interesting characteristics, single stock versus index. And once it was unwound in its relatively smaller size, it was gone. I mean, the risk is gone. And so there were losses that were taken. There were desks that were damaged, but it really wasn't a systemic event in the same way that say even a 2018 type event would have been, certainly not in 1987.