“A man cannot always restrain his own doings and keep them within the limits which he had himself planned for them.”
— Anthony Trollope, The Way We Live Now
Risk Management in Crypto
In the middle of a trading day in about 2007, the risk manager for the equities floor I worked on came around and asked me to put more trades on.
This was a surprising request as my book was already limit-up (i.e., I was taking as much risk as I was allowed to take), but before I could respond, he said, "Yeah, I know. The floor isn’t using all its limits, and if we don’t use them, they’ll give them to someone else.”
I didn’t particularly want to add more risk, but you’d never admit that on a trading floor, so I said, OK, great, and put on the lowest-risk trade I could think of to inflate my book.
(I was a rare breed of risk-averse prop trader, which is like being a skydiver that’s afraid of heights.)
As much as it went against my nature, I understood that using ALL of your risk limits AT ALL TIMES was an institutional imperative in 2007.
Which is a large part of the reason we crashed in 2008.
After reading a number of post-mortems on the 3AC debacle, I’d argue that’s also a large reason why crypto crashed in 2022.
All Together Now
The most surprising thing I’ve learned about crypto recently is how reliant the industry seems to have been on profits coming from the Grayscale Bitcoin Trust (GBTC).
Retail enthusiasm for GBTC caused it to trade at a substantial premium to NAV for three full years, offering a pseudo-arbitrage to the select hedge funds who were allowed to deposit bitcoin with Grayscale in return for GBTC worth 20% or 30% more.
It wasn’t exactly free money — they had to wait six months to sell the GBTC — but as long as the premium held up, it was close enough that it must have felt that way.
That near-free money wasn’t sufficient for funds such as 3AC, however.
Rather than settle for 40%+ annualized returns, they reportedly levered their books up further by borrowing USD against the GBTC they held, swapping the USD for UST, and depositing the UST in Anchor to collect the 19% yield on offer there.
Instead of hedging their GBTC risk, they were amplifying it.
They may not have looked at it that way, thinking the risks involved with GBTC and UST were uncorrelated.
But the pseudo-arbitrage in GBTC and the 19% yield on Anchor were both products of the same phenomenon: retail demand for crypto.
And when the professional leverage piled on top of that retail demand got to be too much, both trades faltered.
The professionals couldn’t help themselves: As with investment banks in 2007, crypto funds were going to take as much risk as lenders would let them.
Celsius, Voyager, BlockFi and others let them take quite a lot, and the positions crypto funds were therefore able to finance proved to be too big for the market to bear.
Voyager, for example, is said to have lent to 3AC on an uncollateralized basis. Not undercollateralized — just uncollateralized.
3AC reportedly borrowed from Voyager at 12% (against zero collateral) and deposited those funds in Anchor at 19%.
(I don’t know what the risk managers at Voyager were thinking, but that is Evel-Knievel-jumping-the-Grand-Canyon-on-a-motorcycle-level risk they were approving there.)
When you can finance a trade without putting up any collateral, the return on that trade is effectively infinity.
Or, in this case, negative infinity.
TradFi does dumb things like that, too, with 2008/9 being the (sub)prime example.
But CDOs were at least complicated — there was nothing complicated about the GBTC or UST trades that have sunk the crypto industry this year.
So here’s the next parallel to the Great Financial Crisis: In 2008, everyone knew there was a housing bubble, but hardly anyone knew how severely the financial system was levered into it.
No one, for example, thought real estate prices in Phoenix, Arizona, would sink the German banking system like it did.
The risk that GBTC would fall below NAV or that UST would depeg was even more obvious than the subprime risk in 2008.
But, as with subprime, I don’t think anyone was aware of how levered long the entire crypto ecosystem was to those two trades.
Consider that 3AC, while borrowing against GBTC to buy UST, was also buying LUNA — the token that was meant to backstop the value of…UST.
That’s like buying houses in Phoenix to hedge your longs in subprime CDOs.
It’s human nature that drives traders to take wacky risks like that when things are going well — no one likes watching other people make money, and in a bull market, the only way to keep up with the competition is to be limit-up.
Even if, like me, that’s not in your nature, you still have to go along with it: You otherwise risk losing your seat as a prop trader and losing your investors as a hedge fund.
These human and institutional flaws are amplified in crypto, because the one thing that DeFi has genuinely accomplished is that it’s perfected the art of recursive leverage.
That should make risk management even more paramount in crypto than it is in TradFi.
But judging from the press reports on 3AC and its lenders, there seems to have been little if any of it being practiced.
TradFi learned its lesson in 2008/9 and, as a result, has been far less levered in the current bear market.
Let’s hope we’ll be able to say the same about crypto in the next one.