Introducing DigFin‘s opinionated weekly briefing of the hottest trends in digital finance. Jumpstart your Monday with something thought-provoking and maybe even amusing. This week: Bitcoin’s back! Our take on crypto spring – and on what we think the lawsuit over a badly performing investment-A.I. is really about.
Unstable coins
Is crypto winter over? The price of bitcoin has surged through $8,000 and, despite a wee pullback, there’s excited talk of it quickly reaching $10,000. Other major coins, including ether and XRP (Ripple’s token) have also enjoyed more recent rallies.
Why now? It could be because digital assets are seen as the new safe haven, at a time when both the U.S. and China are locked in what looks like a long-term slog. The institutional bid is another driver. Investors just need a handful of use-case stories to bet on, and regulated digital tokens could fit the bill.
But the price rises also reflect investors fleeing Tether, a stablecoin related to a dodgy exchange, Bitfinex. This highlights a major problem in the world of digital assets: stablecoins are simply unproven.
The idea of stablecoins is flawed. Just ask any emerging-markets central bank that has suffered a run on its reserves by foreign hedge funds. Or the Bank of England when Soros knocked it out of the ERM. Or the HKMA – Hong Kong! – when short sellers almost made it crash out of its dollar peg in 1998.
And these were major institutions with lots of reserves. A determined short against even well managed stablecoins, ants in comparison to elephantine central banks, can force a collapse of regime. But so far the most prominent stablecoin, Tether, is a farce, with no credibility with its pledge to maintain a one-for-one U.S. dollar reserve.
The rally in crypto is welcome but the institutional framework for digital assets remains very immature and the market too vulnerable to manipulation. That’s the case no matter what happens in the world of fiat money.
Man versus K1: Samathur Li sues Raffaele Costa
A lawsuit alleging a lousy A.I. investment tool has people excited about the prospect of artificial intelligence in the dock. Is that what’s really going on?
Last week Bloomberg reported that an unhappy Hong Kong investor is suing the London-based investment firm that sold him blue-sky views about its whiz-bang A.I.-driven portfolio strategy. The portfolio, of course, tanked.
This may be the first court case over the performance of an investment portfolio driven by artificial intelligence. But what’s actually on trial: the A.I., or the selling practices of Raffaele Costa, CEO and founder of Tyndaris Investments?
DigFin has not met an A.I. specialist yet who would claim their algos ready for hands-off investing prime time. So it’s not surprising to hear that, according to the filing of the plaintiff (Hong Kong real-estate owner Samathur Li Kin-kan), an A.I. called K1 meant to read market sentiment in real time, from unstructured data like social media feeds, lost money.
Li was impressed by Costa’s demos of K1’s amazing performance, so he not only gave him money, money that was geared. And then watched his losses hit $20 million or so.
Perhaps K1’s losses came from the data inputs, or from the trading techniques – apparently the A.I. lacked a human trader’s common sense of timing.
But what’s on trial is not whether K1 performed, or how it worked. What’s on trial is what Costa promised Li the machine could do. The allegation is about mis-selling. Salespeople can jazz up the backtested numbers of the most plodding, analog portfolio. You don’t need A.I. to do that.
Therefore it’s not clear that this court case, though of great interest, is going to settle more existential matters around A.I., such as opening up the black box of its decision-making, or “explainability”. Which is too bad. We need a nice, non-systemic screw-up to parse so that we develop some standards around robot-driven hedge funds before they’re allowed to really cause trouble.
