A nice summary of Li, risk copula and, mostly, the role it had in the crash.
However, critic that I am of associating anything crypto with the world of traceable value, this is where things go south:
“ But if the financial world, couldn’t be mapped with a simple elegant correlation formula, what more cryptocurrencies which suffer from both limited data and are unconstrained?”
This is a serious fallacious “guilt by association”. The “financial world” could never be “mapped” by one simple elegant formula. There are so many different kinds of risk and pricing models for different instruments, durations, and purposes. A large number of these were performing quite well and, as the “civilian” world knows from Michael Lewis, those models worked fine, they were just ignored. Or used Li’s.
The core problem of the “USE” of Li’s model was that it detached entirely from a fundamental data lineage and used purely parametric data to derive risk. So it suffered the fate that any quantitative analysis does: if there is not a clear and direct lineage to some sort of fundamental data, you are in purely speculative territory. When I was at ML during those fun times and ran pricing and trade routing tech for the credit (and rates) desks, there were not a few 50 to 200 page analyses of this sort that were expected to price securities.
The usefulness of these sorts of analyses is variously indicative and should never be used as the only and certain bases for pricing and risk.
Even the holy Black-Scholes model had its moment in the dark when Myron Scholes got to see up close the limitations of these models when he was at LTCM.
But the real lesson here for crypto is, there is no fundamental data. Not even of the fairly abstract version that exists for national currencies. Therefore any attempt to apply any analytic to crypto pricing/risk behavior, except the purely self-referencing parametric, is impossible. Or, at least, until it is regulated like ordinary markets so we can see who is pumping and dumping and why.
You can therefore only call crypto “investing” gambling, as long as you restrict that definition of gambling to the kind of “Numbers” games in the Bensonhurst Brooklyn of old: you don’t get the benefit of the probabilities provided by the legal versions.
This, of course, differs somewhat when it comes to Ethereum or other crypto based off of a transactional usage, such as “gas” or rewards systems. But even these have the problem of the absence of “gas” or rewards usage data. Etc.
A worse situation than betting with Louie the numbers guy.
