FTX didn't hedge its tail risk
When correlations converge on 1, bad things can happen
Nassim Taleb tweeted something yesterday about where he and Karl Popper agree:
I thought this was an interesting tweet, given the recent crash of FTX. FTX imploded, in part, because of correlations: all of its assets started to decline at once. Correlations converged on 1.If all of your long positions suddenly crash in price: you’re fucked. It would have been one thing had FTX been SBF’s personal portfolio of crypto assets. But as we know, it was not: it held client funds. And when clients tried to withdraw funds in a panic, FTX was unable to meet those withdrawal demands because the crash in crypto prices made its balance sheet insolvent.
As I’ve written elsewhere, the implosion of FTX arose because of a solvency crisis, not a liquidity criss. This is fairly straightforward: when all of the crypto that FTX held as reserves crashed in price, it made its liabilities (i.e., client funds) dwarf its reserves. This is a classic case of insolvency, and it meant that FTX could not process all of its clients’ withdrawal requests. Clients couldn’t get their cash out because the assets that FTX held on hand in order to process such withdrawal requests suddenly were worth a lot less than they had been just a few days prior.
FTX failed to consider tail risk, and so had no hedge against it. Here’s Sam Bankman-Fried (SBF) talking about this in his recent interview with Andrew Ross Sorkin:
SBF: In Alameda's case, I don't think I was marking them the way I wish I had from a risk perspective. And I want to sort of differentiate your, like, expected value or, or sort of, like, worth or something like that from security. And, you know, I think that I don't have any strong statements to make about, you know, what value they're assigned from sort of, like, an upside perspective or even a median case perspective. But clearly, I was, I was not nearly cautious enough from a downside perspective – from the extreme downside perspective. And, you know, I can tell you, in my head, I was looking at a 30% down move over a few-day period as a sort of, like, extreme tail-case event that, you know, we had seen once before, and, and then, you know, what happened here was, I mean, a 95% down move over the course of a year, and a, you know, 60% down move over a few-day period with very little liquidity and all happening at once in all of these coins in a correlated fashion in which hedges didn't mean as much also because this was a specific crash on assets associated with Alameda Research, rather than all assets. And so even correlated hedges had limited use there, and a run on the bank at the same time. And all that are things, in retrospect, I should have expected might happen in an extreme scenario, because that's how markets work. And, you know, we've seen other examples of that in history where, when things get really bad, they get really bad for all the relevant things at once, in a very direct and correlated and quick way.
Ex post facto analysis is always easier than ex ante analysis, I suppose, but SBF pretty much admits here that he did not consider the possibility that all of FTX’s positions could decline at the same time. While that’s a tail risk, it’s still a risk, and it is, per Taleb, a risk that ought to have been hedged.