How Our Quantitative Investment Process Adapts to Changing Fundamental Data

The sudden shutdown of key components of the global economy in the COVID-19 crisis has created a challenge for investors, as fundamental data has taken time to respond to the abrupt change in economic activity. In this commentary, we’ll focus on how our quantitative investment process adapts, by design, to scenarios in which data has become stale or backward-looking.

The price-to-earnings (PE) ratio, a key measure of valuation derived from historic annual earnings, is an especially notable instance of a lagging fundamental metric. Since February, airlines have seen trailing PE multiples drop from around 11 to less than 6, because prices have fallen by about 45% on average and reported earnings have not changed. Even forward PE ratios for airlines started at around 9 in February and dropped to about 6.5 in mid-April. Since then, forecasts of 2020 earnings have started to adjust downward, sending the forward PE ratio upward to end April at around 45.

So are airlines cheap, or are they expensive? If only answering that question were as simple as using a PE ratio. Over long periods of time, and applied to hundreds of stocks, building portfolios based on PE alone can add substantial value, but an emphasis must be placed on both the length of time and the breadth of application. In the case of airlines, those that survive the crisis without being nationalized or going bankrupt will undoubtedly be a great investment, but a lot more information is required to identify which airlines are most likely to make it.

This is where we start to peel back the onion of forecasting stock returns, going beyond PE ratios to consider other, more nuanced measures of valuation, and combining these with an assessment of company quality that includes balance sheet strength (e.g., solvency, operating efficiency, liquidity, leverage), sustainability, asset growth, and more. In the case of airlines, stronger companies will be more likely to survive the earnings downturn than those facing debt levels that will prove to be unsustainable in the near term. Unfortunately, getting accurate data on the state of a company’s balance sheet can be somewhat backward-looking, too.

We seek to account for current and forward-looking information that may be benefiting or undermining a company by gauging market sentiment toward stocks. This can take the form of measuring how prices are changing; how forecasts of earnings, sales, and dividends are changing; how hedge funds are changing their positioning; inferring changes in a company by assessing changes in information about customers and suppliers; and examining how company managers are talking about the company in their earnings calls. These are all near-term drivers of a stock’s performance, with their own horizon for effectiveness in an investment process. In the case of airlines, the fastest changing sentiment indicators have been those connected to the downward adjustment of sell-side analyst forecasts, but all sentiment indicators have dropped quickly in that industry group.

We find that when such dramatic changes happen to a company’s outlook, the change of market sentiment has more impact on our forecast return than changes in valuation or quality measures.

We are confident that a combination of longer-term value and quality measures and shorter term measures of market sentiment are ideal for navigating all sorts of economic and market environments, including the current crisis. At the same time, we constantly stress test our process as we manage our portfolios. In the current crisis we are using data and modelling to investigate outsized risk exposures that we may not have captured historically.

One example of this occurred early on in the crisis, as the virus spread through China. In that case, we created a metric to check portfolios for outsized exposure to companies impacted by a shutdown of the supply chain in China, using a data set to map the complex linkages between suppliers and customers across the world. We also systematically monitored the language used in company earnings briefings to look for references to the coronavirus using natural language processing.

As the virus spread to the rest of the world and began to threaten the financial system, we cast an additional risk lens over portfolios to measure exposure to credit risk, illiquidity, excessive debt, and probability of credit default, particularly for companies whose predominant attractive characteristics were value metrics. This was most relevant when we looked at some potential positions in the REIT sector.

Within Real Estate, retail REITs have become a lot cheaper on valuation, but also suffered a very large decline in sentiment. Industrial REITs, by contrast, have become a lot more expensive as a group, with greatly improving sentiment. In general, we see some good contrarian opportunities in
retail REITs, if care is taken to diversify the exposure and overlay additional scrutiny with regard to credit or debt risk. Other specialized REITs dealing with storage or data centers also offer attractive opportunities.

Right now, we are seeing the strongest market sentiment in Pharma and Biotech, Software and Services, Utilities, and Health Care Equipment and Services. We see the weakest sentiment in Autos, Banks, Transport, Energy, and Insurance.

The biggest changes in sentiment (either up or down) have occurred in the following industry groups across global developed market equities:

Financial stocks now appear cheaper than they did at the start of the year, but our overall forecasts have not improved for financial firms because sentiment has declined so dramatically. We still prefer insurers (where sentiment has declined, but less steeply) over banks.

By incorporating short-term sentiment signals, in addition to longer-term value and quality metrics, our investment process is designed to adapt to rapidly changing market conditions. In addition, we consistently stress test our portfolios on an ongoing basis, which also adds a dimension of adaptability to our process in ordinary as well as in distressed market environments.