Lessons from Quantmageddon

A review of portfolio construction could help investors who got burned tracking the ‘Momentum’ factor in 2019.

Momentum investors took a massive hit in September 2019, which analysts at Nordic bank SEB dubbed ‘Quantmageddon’. Momentum is one factor-based strategy that has proven popular with investors in recent years [see box: Commonly used factors for investing], and equates to buying into stocks that are rising – it went into reverse from mid-August (see Fig 1).

On 13th September, Morgan Stanley analysts were reporting Momentum was down 12% over the week and 19% since its high on 26th August. Alix Guerrini, equity quantitative strategist at Morgan Stanley analyst noted 2.5% daily drops on both 5th and 9th September (see Fig 2). This followed the strategy’s best month since June 2008, according to Goldman Sachs analysts.

The inclusion of factors within a fund can be expressed at different levels of sophistication, using anything from quantitative models in mutual and hedge funds, to a weighted index for an exchange-traded fund. That has a material impact on the returns and risks of funds.

Vitali Kalesnik, partner and senior member of the investment team at Research Affiliates, says, “The most vulnerable are investors in long-short leveraged strategies.”

Consequently, the longer-term impact of the 2019 reversal varied by investor with sources reporting significant losses to many hedge funds. Morgan Stanley’s analyst team reported “many investors do not plan to change their overall portfolio structure and view this as a technical unwind.”

However, there are potentially lessons that can be learned from the event. Some market observers report the event was entirely predictable, leading to questions around how investors can mitigate future risks, while still tapping the potential returns that factor investing offers.

Warning signs

To assess the possible risk mitigators, it is necessary to consider how factors and particularly Momentum relate to the rest of the market.

“The industry has fallen into this approach, where typically four or five factors are viewed as risk premia,” says Scott Chronert, analyst at Citi. “Over longer time frames, they are presumed to independently give you an outperformance dynamic, so time frames become relevant.”

Within investment management, the approach to using factors has morphed into long-only products with a stock-selection approach that aligns with a factor definition, and more classic quant approaches where the factor is represented by going long on the top quintile or decile of securities and short on the bottom quintile or decile.

“For long-only factors we think they do end up being composed of sector bias and stock bias,” says Chronert. He observes that as rates were falling through the summer months, with increasing concern about global growth due to trade concerns, this impacted factors.

“The falling rate trend was taking on, in our view of it, almost bubble-like characteristics in the broader market, to the point where even momentum was taking on a falling rate bias. Momentum in the market for much of last year was a falling rate beneficiary, which tends to align with the more Minimum Volatility (Min Vol) approaches,” he says.

Min Vol tends to be most negatively correlated to interest rates, while Value tends to be the main factor that has a positive rate correlation bias to rates. Consequently, the falling rates and rising trade worries was to the benefit of Min Vol investors, and to Momentum which was taking on the falling rate dynamic.

“Walking into early September, suddenly a very solid employment report turned that falling rate thesis around growth concern, upside down,” Chronert says. “Very quickly that triggered a rally in the types of sectors and stock that would be correlated to rising rates, and a sell-off in the consensus trade beneficiaries of a falling rate.”

To manage the risk of volatility like this, a rapid about-face, following a sustained period of growth, the likelihood and timeliness of such an event have to be considered by investors.

Crashes are to be expected

In the February 2019 paper, ‘Alice’s Adventures in Factorland: Three Blunders That Plague Factor Investing’, analysts from Research Affiliates observed that amongst the risks that investors seemed to be blind to, in their use of factors as the basis of an investment strategy, “returns falling far short of expectations due to overfitting and/or crowding, drawdowns that far exceed expectations, and failed diversification, as correlations unexpectedly soar.”

Kalesnik says, “Individual factors are prone to sharp drawdowns and periods of underperformance. This is a risk of factor performance, it is an unavoidable feature of factor investing and if investors are not prepared for that, they should not be investing in factors.”

He observes that there are positive and negative effects as a result of this behaviour.

“The bad thing is the risk,” he says. “The good thing is that because factors have this risk profile, they are too risky to be held by everyone. Investors may think they are comfortable tolerating the risks but after seeing it underperforming for three years it turns out they are not.”

Chronert notes that investors need to understand way that macro elements such as rate changes are connected to the way factors will trade in the marketplace.

“You set yourself up for these reversal circumstances when there is an underlying trend around a macro which plays into how the factors are trading, and then something happens that changes the sentiment around that macro,” he says. “In this case if you look at the 10-year US treasury trend over the last year and overlay it with how a couple of these factors traded, it lines up pretty closely.”

There also are quantitative indicators which some market participants cite as potential risk warnings. According to data from FactSet/Morgan Stanley research, the Z-score – indicating standard deviations (SDs) from the mean – of Momentum was over 2.7 SDs in August, which Morgan Stanley’s analysts noted “reached overbought levels” which they said, “has previously been a very effective warning sign of a forthcoming reversal in the performance of momentum factors.”

Risk management

Matching the investment strategy to one’s risk appetite can only be achieved if the strategy is fully understood. In factor investing that can be complicated. Firstly, investors need to read between the lines when they are sold the strategy.

“Factor investing got a little too much positive attention in the sense that it was over-marketed. To a degree, investors were promised unrealistic returns,” says Kalesnik.

Secondly, any factor can be expressed in different ways within a strategy based on the sophistication of the model used. That has a direct impact on the risks that a portfolio might carry in the event.

“The simplest way of building the factor is to take one indicator. For Momentum, the prior 12-month returns,” says Raul Leote de Carvalho, deputy head of the Quant Research Group, at BNP Paribas Asset Management, which issues factor-based funds. “If you use that to select stocks of equal weight you have a number of problems, because you catch many other risk exposures which you don’t want to have. You may have beta which is changing all over the place, you have biases such as market cap, which impact returns but are not what you are aiming at. You want to extract the Momentum premium, and you have to correct for bias so you get the premium for the style in the purest way.”

To adjust for beta effectively, the investment manager will need to have forecast beta for each stock. The accuracy of that hinges upon the risk models that the asset manager can support.

Gregory Taïeb, quantitative investment specialist at BNP Paribas Asset Management, says, “By purifying Momentum, by reducing its dependence on risk factors like beta [the security’s correlation to the benchmark], you increase the information ratio you can extract from Momentum style; you can reduce the correlation between Value and Momentum.”

Leote de Carvalho says, “We calculate the beta for the stock and we correct the return of the stock for the market exposure and we work only with idiosyncratic momentum – stock beta is less accurate than portfolio beta but still mitigates the very strong crashes in Momentum. It doesn’t remove them completely but reduces them.”

Finally, the investor must be cognisant of the likelihood for crashes within an investment window, and to have a realistic timeframe in which to target returns.

“There are macro, sector and individual stock influences that determine how the factors perform in the marketplace in the shorter term,” says Chronert. “Min Vol might outperform over longer time frames by going down less in a recession. Value on the other hand might perform better over longer time frames because it gets in early coming out of a recession. So the time and horizon that’s being considered is the crux of all of this. We think you have to pay attention both to the macro influences and some of the fundamental influences on some of these factors to help determine how you want to position over a shorter time frame.” ●

TOP OF PAGE