For hedge funds November, with its post-election market swings and the Jensen Huang affair, has so far been bruising. October, though, was brutal: many funds had their worst month since the covid-19 pandemic began, according to JPMorgan Chase, America’s biggest bank. A wobble in markets’ enthusiasm for artificial intelligence forced a sell-off of technology stocks. The surprise re-election of Brazil’s left-wing president, worries over France’s sovereign debt and tumult in credit markets added to the pain. A Bloomberg index of the industry’s returns fell by 1.7% in the month. The biggest fund at Millennium, a giant manager, lost 2.3%. The main fund of Arini Capital Management, a fast-growing credit shop, was down by 19.7%.
By the time tech stocks plunged again on November 5th, managers were already so light they had little left to sell. According to data compiled by UBS, another big bank, there were six days in the previous month on which funds had unloaded more. Hedge funds sometimes get the blame for market falls. This time they were innocent, though they have been given food for thought. The industry has been coasting on the AI boom: the Bloomberg returns index still ended October 4.1% higher then it began the year. But October exposed the dangers of being caught on the way down.
One category, multi-strategy funds, has more to ponder than most. Whereas a typical hedge fund implements the vision of a single portfolio manager or a few working together, multi-strategy funds recruit several, all of whom are free to pursue their own trades. The central fund constrains them only by controlling capital allocation among its various “pods” of traders and imposing a strict risk policy to limit losses. In theory the managers’ independence makes their trades unrelated, raising the whole fund’s risk-adjusted performance compared with that of one manager. In booming markets returns are also less volatile than in single-manager funds.
Multi-strategy funds were pioneered by such luminaries as Ken Griffin of Citadel, which he founded in 1990, and Izzy Englander of Millennium. Others have enjoyed remarkable success, too. PivotalPath, a consultancy, estimates that the ten largest control gross assets worth $2.2trn. But in recent years their rise has levelled off—and in recent weeks the tight risk policies have begun to pinch. Many force managers to start unwinding positions when losses exceed only 5%. Lose more than 7.5%, and a pod may be shut down. With leverage, even small movements are enough to derail trades.
October’s movements were more than enough to force liquidations. By November 9th, when investors had cooled on the financial instruments enabling the buildout of data centres by technology giants from Nvidia to Meta, 32% of hedge funds’ books were exposed to AI, down from 47% in June.
Managers became timid. Funds, which borrow from markets as well as using their investors’ cash, reduced their net debt by 8% of their net exposure in the week to November 6th. This may reflect more than caution: it has been a bumper year for attracting new investment, the best since 2007. But mindful of scaring investors away, fund managers are also eager to limit damage to their returns with too much deleveraging.
The episode raises questions for the multi-strategy model. Its supposed advantage is the managers’ independence. With several supposedly making insightful independent decisions, the fund should be protected should some calls not pay off. Gains from a trade on steel, say, could cushion losses from a bet on neocloud firms. But it is increasingly difficult to avoid AI. In July a plunge in chipmakers’ shares forced funds to sell stocks and bonds whose value depended on the technology’s success. Burnt, many managers promised not to load up again. Just a few months later, they must have déjà vu.
There are two reasons for this. The first is the sheer amount of the financial system now within the reach of multi-managers. Finding a broad enough range of instruments with which to deploy $2trn-worth of firepower is hard even in normal times. The second is that now almost no corner of the world economy is untouched by AI. A portfolio manager buying copper futures is of little use to balance the risks taken by a colleague backing neoclouds if the price of copper depends on AI’s future development. The growing pervasiveness of one technology may ensure the multi-strategy hedge fund’s best days are behind it.