Quantitative investing is generally defined as the use of a rigorous set of rules based on easily observable criteria to guide investment decisions. It encompasses a wide variety of strategies, from longer-term equity strategies where equities are bought and held for periods up to or longer than a year, to very short-term strategies trading in and out of securities multiple times during the course of a day. Of course, these shorter-term strategies do not really “invest” in the securities. Regardless of the exact type of strategy employed, there are several common steps in all types of quantitative investing: (1) screens, (2) backtests, and (3) implementation. This general process is graphically represented in Figure 1.1.
Figure 1.1 Life cycle of a quant investment
A screen is a formula for deciding what to buy and sell at any given time. A simple screen might be: Buy all stocks that have positive returns over the last year.1 This is an example of a stock screen, which divides the universe of stocks into those the system will pick to buy and those it will not. A more complicated stock screen could have multiple criteria, such as “buy all stocks that have positive returns over the last year AND price to earnings (PE) ratios between 5 and 15.”
As opposed to stock screens, you can also implement timing screens. Timing screens are screens that use some measure of the overall economy to decide whether it is a good time to buy or sell. An example of a timing screen could be “Move all assets to SPY if unemployment has fallen two months in a row. Otherwise, keep all assets in cash.” SPY is the ticker for a security that mirrors the S&P 500 index. Thus, the screen basically keeps money in the broad equity markets if the economy is stable and in cash if unemployment is rising.
The difference between timing screens and stock screens can be subtle, and a number of screens will have flavors of both types. “Buy all stocks if unemployment has fallen two months in a row, and buy only stocks with PE ratios between 5 and 15 if not” is a screen that combines timing and stock screen elements. This book generally keeps them separate, but it is important to remember that they could be used together, if desired.
A screen generally returns a list of stocks and then a decision must be made. For example, the screen “Buy all tech stocks with PE ratios less than 10 that are profitable and pay a dividend” might return a list of about 15 stocks. The next step is to decide what to do with these lists. Generally, the easiest (and most popular) strategy among quantitative investors is to invest capital equally in all stocks that pass through a screen. So if 15 stocks pass through the screen, you divide the money you are investing equally into the 15 stocks and that will be your portfolio. There are other ways to weight the list: For example, you can weight by market capitalization or by dividends. Alternatively, you could do deeper fundamental analysis on the company names generated by the screen and pick individual securities to invest in, rather than simply dividing your capital equally among the companies that pass through the screens.
You should feel free to use screens to support deeper analysis of individual companies. Chapter 4, “For the Deeper Divers among Us: How to Use Quantitative Strategies to Enhance Fundamental Valuations,” explores two case studies that do just that. However, this book does not cover techniques to do deep analysis into individual companies.2
Horizons (or Rebalancing Frequency)
A key part of any quantitative investing screen is the horizon of the strategy. If the screen provides information that determines what the trade will be, the horizon provides the duration over which the information in the screen is valid. After the horizon is reached, the old positions are closed out, and if desired, the screen can be run again and new positions chosen. This process of closing out old positions and choosing new ones is known as rebalancing. The horizon of your strategy is, thus, also known as the rebalancing period. Rebalancing is a very important part of quantitative investing and Chapter 8, “RebalancingWhy, How, and How Often,” is dedicated to this concept.
Many times, the measures used in the screen dictate the horizon. So, for example, in the screen discussed earlier that was based on unemployment, the strategy must have a horizon of at least one month because unemployment numbers are released monthly. Put a different way, horizons cannot be shorter than the frequency at which information is released (if they were, the stocks in the screen would remain the same until new information came out). The screens based on PE ratios can have a much shorter horizon because the price of a stock is constantly changing and, thus, PE ratios are constantly changing.
If the trading horizon is longer than the information frequency, you risk that the information previously used to generate the portfolio has gone stale in between rebalances. For example, you sort on PE ratios today and buy all stocks with PE between 0 and 10. Tomorrow, one of the stocks you bought spikes, its PE is more than 10 and no longer passes through your screen. If your horizon is longer than a day, you will hold this stock despite the fact that the stock no longer passes through your screen. Thus, the shorter the horizon, the less likely stocks that are in your portfolio that do not pass through your screen, and the less likely that your portfolio excludes stocks that have evolved to pass through your screen.
The other main factor driving strategy horizon is transaction costs. The shorter the horizon of your strategy, the more times you will have to trade out of old positions and into new ones. Trading generally incurs transactions costs, both in terms of commissions, price impact when trading, and time and headache spent executing your trades.
This is the main tension in choosing the ideal strategy horizon. Too long and youre holding stocks based on old information; too short and your returns are being eroded by transaction costs. In the financial industry, trading horizons can range from the extremely long (a year or even longer) to the very, very short (microseconds or even nanoseconds). High-frequency trading is the form of quantitative “investing” that deals with these very short holding periods. However, high-frequency trading really is not “investing” because positions are rarely held overnight and rarely is the intention of such trades to actually invest in companies. This book, for most strategies, runs a quarterly rebalance. Equities Lab (the tool that most of the analysis in this book uses) has a default rebalancing period of a week, although that can be easily changed.