Quantitative trading strategies have a reputation for being difficult to explain and even harder to understand. In his new book, Inside the Black Box, Rishi K. Narang provides a first-hand, plain English perspective on quant trading.
How it typically works
The best way to understand both quants and their black boxes is to examine the components of a quant trading system. The diagram shows a schematic of a typical quantitative trading system. It portrays the components of a live, "production" trading strategy (e.g., the components that decide which securities to buy and sell, how much and when) but does not include everything necessary to create the strategy in the first place (e.g., research tools for designing a trading system).
The trading system has three modules – an alpha model, a risk model and a transaction cost model – which feed a portfolio construction model, which in turn interacts with the execution model. The alpha model is designed to predict the future of the instruments the quant wants to consider trading for the purpose of generating returns. For example, in a trend-following strategy in S&P 500 futures, the alpha model is designed to forecast the direction of the S&P 500 futures markets.
Risk models, by contrast, are designed to help limit the amount of exposure the quant has to those factors that are unlikely to generate returns but could drive losses. For example, a trend follower could choose to limit his directional exposure to a given asset class, such as commodities, because of concerns that too many forecasts he follows could line up in the same direction, leading to excess risk; the risk model would contain the levels for these commodity exposure limits.
The transaction cost model is used to help determine the cost of whatever trades are needed to migrate from the current portfolio to whatever new portfolio is desirable to the portfolio construction model. Almost any trading transaction costs money, whether the trader expects to profit greatly or a little from the trade. Staying with the example of the trend follower, if a trend is expected to be small and last only a short while, the transaction cost model might indicate that the cost of entering and exiting the trade is greater than that expected profits from the trend.
The alpha, risk and transaction cost models then feed into a portfolio construction model, which balances the trade offs presented by the pursuit of profits, the limiting of risk and the costs associated with both, thereby determining the best portfolio to hold. Having made this determination, the system can compare the current portfolio to the target portfolio, with the difference between the current portfolio and target portfolio representing the trades that need to be executed.
The goal of a portfolio construction model is to determine what portfolio the quant wants to own. The model acts like an arbitrator, hearing the arguments of the optimist (alpha model), the pessimist (risk model) and the cost-conscious accountant (transaction cost model) and then making a decision about how to proceed. The decision to allocate this or that amount to the various holdings in a portfolio is mostly based on a balancing of considerations of expected return, risk and transaction costs. Too much emphasis on the risk can lead to underperformance by ignoring the opportunity. Overestimated transaction costs can lead to paralysis, because the trader may end up holding positions indefinitely instead of taking on the cost of refreshing the portfolio.
Quantitative portfolio construction models come in two major forms. The first family is rule based. Rule-based portfolio construction models are based on heuristics defined by the quant trader and can be exceedingly simple or rather complex. The heuristics that are used are generally rules that are derived from human experience, such as by trial and error. The simplest example of a rule-based portfolio construction model is to target an equal weighting of all the positions in the portfolio.
The second family of quantitative portfolio construction models is based on optimization. Optimizers utilize algorithms – step-by-step sets of rules designed to get the user from a starting point to a desired ending point – to seek the best way to reach a goal that the quant defines. This goal is known to be an objective function, and the canonical example of an objective function for an optimizer is to seek the portfolio that generates the highest possible return for a unit of risk. By their nature, optimizers can be more difficult to understand at a great level of detail, but they are straightforward conceptually.
The importance of data
It is difficult to overstate the importance of data, and it can be seen from many perspectives. First, data are the inputs to quant trading systems. It turns out that the nature of the inputs to a system dictates what you can do with the system itself. For example, if you were handed a lot of lettuce, tomatoes and cucumbers, it would be difficult to build, say, a jet engine. Instead, you might decide that these inputs are best suited for making a salad. To make a jet engine, you more or less need jet engine parts, or at least materials that can handle high velocities and acceleration, high altitude and a wide range of temperatures. The same is true with quant systems. To the extent that you are given data that focus on macroeconomic activity, it is extremely difficult to build a useful model that does not somehow reflect macroeconomic concepts.
Frequently, many details of the model itself are driven by characteristics of the inputs that are used. Refining our example, imagine that you are given slow-moving macroeconomic data, such as quarterly U.S. gross domestic product (GDP) figures; furthermore, you receive them only a week after they are released to the public. In this situation, it is unlikely that you can build a very fast trading model that looks to hold positions for only a few minutes. Furthermore, note that the U.S. data you get might be useful for predicting bonds or currency relationships but they might not be sufficient to build a helpful model of equity markets. U.S. GDP data will also tell you little about what is happening in Uruguay or Poland in any of their securities markets.
Theory versus data
The distinction between theoretical and empirical science is germane to quantitative trading in that there are also two kinds of quant traders. The first, and by far the most common, are theory driven. They start with observations of the markets, think of a generalized theory that could explain the observed behavior, then rigorously test it with market data to see if the theory is shown to be either false or supported by the outcome of the test. In quant trading, most of these theories are things that would make sense to you or me and that seem sensible when explained to friends at cocktail parties. For example, "cheap stocks outperform expensive stocks" is a theory that many people hold. This explains the existence of countless "value" funds. Once precisely defined, this theory can be tested.
The second kind of scientist, by far in the minority, believes that correctly performed empirical observation and analysis of the data can obviate the need for theory. Such a scientist’s theory, in short, is that there are recognizable patterns in the data that can be detected with careful application of the right techniques. The example of the Human Genome Project is instructive. The scientists in the Human Genome Project did not believe that it was necessary to theorize what genes were responsible for particular human traits. Rather, scientists merely theorized that the relationships between genes and traits can be mapped using statistical techniques, and they proceeded to do exactly that. Empirical scientists are sometimes derisively (and sometimes just as a matter of fact) labeled data miners. They do not especially care if they can name their theories and instead attempt to use data analysis techniques to uncover behaviors in the market that aren’t intuitively obvious.
In all, with a little bit of interest and effort, the average investor should be able to understand not only how quants make money and what risks they take, but also how the trading strategies they utilize really work.
Rishi K. Narang is the founding partner of Telesis Capital and author of Inside the Black Box: The Simple Truth about Quantitative Trading (John Wiley & Sons, 2010).