Driving Investment Decisions With Data

(This article first appeared in the 2020 Q4 issue of GlobalTrading.)

By Richard Lacaille, Global Chief Investment Officer, State Street Global Advisors

 Richard Lacaille, State Street Global Advisors

Data-driven decisions are playing an increasing role in the investment process, with portfolio construction and execution becoming more data-driven and integrated. This has been a consistent theme across listed products like equities for years, and we are seeing an increased adoption in other asset classes. The growth and importance of data doesn’t come without challenges, complexity increases, and the need to continuously evaluate new data’s contribution to alpha and the investment process. One thing we do know is the amount of data available isn’t decreasing, as Bank of America recently noted in a research report: “We will create as much data in the next two days then we have done since the dawn of civilization through to 2000.” (BofA Alternative Data Primer 10-21-20). The key to turning these useful ingredients into alpha, however, is to analyze the context and time horizon where a particular data set can add value and then to construct a befitting investment process from portfolio construction to execution.

Portfolio Construction

Active investment strategies need to have a clear idea about how they are finding and exploiting market inefficiencies, and then have an investment strategy and process, fueled by data. Before the emergence of quantitative investing, a typical process might have involved trying to estimate the fair value of a business based on discounted future cash flows and therefore dependent on estimates of revenues and costs into the future. Quantitative techniques boiled down the problem to smaller number of variables of interest, building on theoretical and empirical research but did not typically include “alternative data,” but also replicated to some extent the process of evaluating what the balance is between the price of a business and its future discounted cash flows using a variety of proxies. Time horizons and turnover rates were not dissimilar between the two styles and they could both be characterized as a sort of valuation arbitrage, seeking outperformance by mis-pricings against an evolving fair value.

A critical element for both styles is the estimation of future earnings and sell-side and buy-side analysts therefore both become avid consumers of data that incrementally improve near term forecasting power and the prevalence of new alternative and alternative data is an important part of this. The difficult question for investors then becomes “is my investment process likely to benefit from the consumption of data that helps me to forecast next quarter’s earnings?” For some the answer might be yes, but it is important to understand that the implication might be an investment process that attempts to profit from the small changes in earnings estimates incrementally implied by the arrival of new data.

The longer-term value of companies is mostly a reflection of the longer term cash flows. However, these are really hard to forecast and not all alternative data is particularly helpful in this regard. Furthermore, conventional active managers should probably stop playing the short-term valuation arbitrage game against those who can assimilate the data and are willing to trade rapidly. Their focus should be on the data with more eclectic and qualitative nature that has the ability to drive longer term value. If you wanted to forecast Apple’s market share, for instance, vs. Samsung’s in five to 10 years, the more important data points might be the strength of the company in R&D, its ability to retain key creative talent by embedding the right culture, etc. together with many other factors like intellectual property rights that are not so easily captured with traditional data. 

Execution

SSGA Trading has long been a proponent of process, technology and data-driven execution. Trading and portfolio managers work closely across asset classes, and underlying the investment process is Transaction Cost Analysis (TCA). Since its start in 1999, our internal TCA contains over 150 million records and currently covers Equities, Futures, Currency and Fixed Income Trading. This data series allows us to monitor our costs and adjust our portfolio construction and execution approaches for different market regimes. Our investment teams and management utilize Tableau software for data visualization to interact with our TCA data and drive better outcomes. 

SSGA Trading has been an early adopter of data-driven execution approaches. Some examples include Execution Management Systems (EMS), Algorithmic (Algo) trading and Algo Wheel technology. EMS is software utilized by traders for connectivity, aggregating data, displaying analytics and creating rules-based automation. Algorithmic trading utilizes pre-programmed trading instructions to account for order and market variables such as size, price, volume, volatility and timing. An Algo Wheel, which we deploy globally for equities and futures, is routing technology that looks at order instruction and routes to a specific algo type with the counterparty chosen based on historical TCA performance. The technology removes trader biases, improves transparency, encourages broker competition and has resulted in lower transaction costs. 

These data-driven approaches aren’t just for listed products, they are expanding to other assets classes as transparency, data and trading protocols evolve. EMSs have long been used in equities and futures. They are expanding to Fixed Income products. Like other asset classes, we are seeing data bring portfolio construction and execution more tightly integrated. For example, in credit, there is a bigger need for sampling and portfolio optimization given the large number of securities and the small percentage that trade daily. Citi notes in 2018 of the 21,175 publicly registered bonds outstanding last year, just 246 or 1% traded daily. (Citi Global Credit View 11-06-19). To build portfolios more efficiently and take advantage of liquidity opportunities real-time, asset managers are bringing trading data upstream into the credit portfolio construction process. 

Conclusion

We expect data-driven decisions will move portfolio construction and execution more closely aligned across asset classes. As the investment process evolves, execution strategy will become more dynamic and will consider portfolio characteristics and historical behavior; end investors will ultimately benefit from this trend.