In factor timing, ‘where?’ matters as much as ‘when?’

Goldman quants’ thought experiment shows timing works best for low-Sharpe strategies

Quants have long tried to pinpoint when the factors they work with are primed to crash or surge. Many have sought ways to ‘time’ these switchbacks. Most agree it is hard; some say it can’t be done.

Perhaps, though, quants have been asking the wrong question.

Rather than asking ‘when?’ to time the factors that drive asset returns, researchers at Goldman Sachs quants asked ‘where?’ timing could work.

Nick Baltas and Desi Ivanova from the bank’s systematic trading strategies research team designed a thought experiment to help answer that question. They found the benefits of timing were more pronounced for strategies with lower Sharpe ratios.

Baltas and Ivanova sketched the distribution of returns for two hypothetical strategies in which a manager is able to time the direction of the investment perfectly, eliminating all losses. The distributions appear as conventional bell-shaped curves but with the left side pressed flat (see graph).

 

It’s clear from comparing them that the higher Sharpe strategy benefits less from timing in relative terms. Intuitively, for the better strategy, there are fewer potential losses to cut out by correctly calling changes in direction.

Of course, returns for real strategies rarely look like a bell curve. Perfect timing would be more valuable for a strategy that loses big every so often. But anticipating such events is hard. And in simulations, the benefits were small, the quants say.

Goldman’s findings on where timing is likely to work best has special relevance today. Several investors are champing at the bit to time the equity value factor – a popular quant factor traded through buying stocks that look cheap on ratios such as book to price – which has been cheapening drastically without any clear fundamental reason.

AQR’s Cliff Asness, one of factor investing’s originators, used to tell clients to increase exposures only when a factor offered a fabulous bargain, and then only in moderation. In November, he told investors that time had come. “Factor timing is an ugly thing. But I think it is about time we did some,” he wrote in a note.

Rob Arnott, another luminary in systematic investing, has said repricing of value stocks could yield a 16% premium.

So, do Goldman’s quants think factor timing is worth a go in real life? Pick your battles, Baltas and Ivanova say.

In their simulations, monthly timing calls on higher Sharpe-ratio, short-volatility carry strategies would need to be around 80% accurate to make timing worthwhile – far beyond what seems reasonable.

But for long-only equity factor strategies, rebalanced monthly, the bank found a more modest hit rate of 60%, more or less, would make timing worthwhile.

The authors call that “borderline realistic”. For comparison, a trend-following signal might call its underlying markets correctly 55% of the time on average.

The Goldman quants’ findings should convince would-be timers to give up on strategies where their efforts are likely to flop. Among equity value investors, meanwhile, 60% will doubtless be good enough odds for some.