Low beta, multi-factor and global approach

“The major contributor to performance was being short high beta stocks, which did very poorly over the year, especially in the US,” explains Harindra de Silva, Co-Head of Research and portfolio manager for the Systematic Edge team at Allspring Global Investments, which applies systematic strategies to equities, fixed income, customised separately managed accounts, multi-asset strategies and options. Within the equities vertical, GLS sits under Allspring’s “defensive” umbrella; the manager also offers equity income, sustainability-oriented, and alpha strategies. 

GLS short performance is typical of its history: since inception in December 2009, the short book has contributed positively to performance despite high single digit annualized returns for the market over the same period.

The strategy’s long book is also lower beta than the market but this should not be conflated with low volatility strategies that are also managed by Allspring. “The structural bias to low beta is the dominant anomaly, and this should not be mixed up with low volatility, which is not an integral part of the strategy,” clarifies de Silva.

GLS performance in 2022 lagged some quant long/short and market neutral strategies that are more value oriented and are wholly or mainly focused on the US. “We are more diversified, with exposure to value, quality, momentum, and even some small caps. We are also globally diversified, and value outperformed by most in the US,” says de Silva. Indeed, Allspring’s US-only mutual fund version of the long/short strategy (ticker ADMQX) somewhat outperformed the global strategy in 2022.

GLS did have material exposure to value, which was most helpful in the first half, while maintaining quality exposure was more additive in the second half of 2022 – and parts of the first quarter of 2023. 

Turnover between and within factors 

The amount of factor rebalancing varies considerably over time. “In March 2023 there was not much change versus the end of 2022, whereas 2022 was one of the most active years for factor turnover,” says de Silva. At a high level, factors can be broadly categorized into financial risk, growth, quality, revisions, technical and valuation. In 2022, the Systematic Edge Global Alpha model was consistently short of financial risk, and over the year it switched between short, neutral and long exposures to growth. The quantum of long exposures to the other four factors fluctuated over the year.

Each of these six factor categories, and a seventh one, liquidity, are further split into sub-factors, which are considered individually. The approach to defining and measuring factors is quite nuanced because they are analysed from many angles. For instance, Allspring’s value exposure is based on nine metrics and their weightings shift over time. “In the first half of 2022 we had a big loading on price to sales, which then came down, while dividend yield increased in weight. These were the two biggest changes within the value factors,” recalls de Silva. Quality also saw some internal turnover amongst its seven metrics: “We have become overweight asset utilisation, because of its importance in a slowing economy”.

These switches of emphasis are based on repeated patterns observed over long periods of history. “We have followed factor momentum and mean reversion over 20 years. Within the value factor, in the early stage of a recovery, price to sales works well, and in a more mature stage of recovery, forward PE works better,” explains de Silva.

Monthly exposure and performance attribution reports break down around 30 granular factor contributors. The weightings of these sub-models are based on a mix of momentum over various lookbacks and mean reversion, which augment or diversify the forecasted factor return depending on agreement.

The weightings vary with the economic climate as well as technical factors. For instance, “We are usually short analyst dispersion, within the financial risk factor category, because investors prefer stable earnings but in a recessionary environment it can become a positive loading,” points out de Silva. 

Directionally factor agnostic 

While some quant funds constrain their preferred factors to a minimum positive weight or zero, Allspring can in theory go long or short of any of its 32-factor metrics, at the aggregate portfolio level. Some factors show zero active exposure in reports, but this does not imply that they have been temporarily switched off or permanently retired. Instead, it simply means there are more attractive factors to allocate capital to.

Overall, the factor exposures are more diversified than for some other quant strategies. In theory, some overriding constraints prevent the strategy from becoming overly concentrated into any one factor. “But in practice we have never needed to action the caps and constrain factors,” says de Silva.

“A further nuance is that headline metrics for the whole book do not tell the full story of what is happening at position level. The portfolio might like the same stock even though our factor forecasts change because that stock’s factor loadings have changed,” says de Silva.  

Subjective sector betas and exposures 

The stock-picking signals are also cognizant of dynamic shifts in beta, which are in turn sensitive to lookback periods and techniques used to estimate beta. 

While a growth investor or a US-biased global investor will often be overweight both technology and healthcare, Allspring’s multi-modal approach can lead to a long in one and a short in the other. In March 2023, the portfolio was underweight technology. “This is mainly because its shorter-term, post-Covid beta has risen (investors using a two- or three-year lookback window, including the Covid period, might instead deem tech to be low beta). Conversely, the beta of healthcare has come down, which helps to explain the overweight stance,” says de Silva. Energy and defence betas have also fallen since Russia’s invasion of Ukraine, and their portfolio weighting has increased. In contrast, regional banks’ beta sharply increased in March 2023 as several banks failed. GLS has recently been neutral to financials and underweight US regional banks. 

Sector exposures are also influenced by other risk constraints, such as an objective to remain neutral with respect to interest rate and inflation sensitivities, which became more important in the second half of 2022. “Low beta stocks such as utilities can have very high interest rate sensitivity, which we might end up with if we did not have this constraint,” says de Silva.

Shorting glamour while sidestepping memes and manias 

Allspring’s short book contains many unprofitable or barely profitable firms, and therefore has an extremely high PE ratio and forecast earnings growth of 23% as of January 2023. De Silva judges that, “The growth rate is too optimistic because very few companies will grow at 23% for three of four years, and the PE ratios assume these firms can grow their earnings unsustainably”. 

Some of these glamorous growth stocks can be costly to short, in terms of borrow and price volatility, but Allspring controls these risks amongst others. “If a stock is really expensive to borrow, it will not be a big position. Generally, high short interest is a really good predictor of borrow costs. We also size inversely to implied volatility and CDS spreads. Moreover, our sentiment engine reads news and may lead us to shrink positions in companies that get more news coverage than they received historically,” says de Silva.

In addition, machine learning is used to identify and reduce short exposure to companies that are not well explained by economic fundamentals and linearly interpreted factors, which can include some firms exposed to cryptocurrencies, AI, and other nascent themes. 

All of these safeguards mean Allspring rarely has any large short position in “meme stocks” which have heavy option trading and social media coverage, though there can be some exceptions. A short in an early-stage electric vehicle was recently profitable, meanwhile Allspring fortunately avoided short exposure to firms like GameStop and AMC. 

Outlook 

The outperformance of higher beta and growth stocks has become a headwind in the first quarter of 2023 but this has also helped to increase valuation dislocations and opportunities. De Silva finds that valuation gaps have converged across industry groups, but not within them. “On an industry basis, differences have narrowed a lot, but on an industry-adjusted basis they are just as large. A lot of the adjustment occurred inter-sector, but there are still big opportunities for both inter- and intra-sector valuation reconvergence inside and outside the US.” This bodes well for the strategy’s suite of valuation factor metrics.