Quantitative trading, or quant trading, is an approach to financial investment using multiple large datasets in conjunction with advanced mathematical techniques and statistical modeling. Computer algorithms are commonly used in order to sift through data, quickly iterating through hundreds of thousands of data points. In contrast to algorithmic trading, quant trading involves a highly computational mathematical approach when deciding appropriate trading conditions whereas algorithmic trading uses traditional technical analysis (i.e. using past data). Quantitative trading has garnered a lot of attention especially in recent years, and constitutes a significant portion of many hedge fund and investment bank’s assets alike.
As mentioned briefly, the main goal of quant trading is to identify and capitalize on appropriate trading conditions to generate consistent profits. In general, these conditions arise from market inefficiencies coming from imperfect information about an asset's true price; in other words, certain assets may be under or overvalued, which can later prove valuable when liquidating a stock holding. To give a more concrete example, when a stock is undervalued, it would be appropriate to buy. When certain conditions are met—which is up to the trader to decide—an asset is bought, sold, or more commonly, nothing is done. Other examples of appropriate trading conditions can be determined using technical indicators covered in more detail here.
While this important aspect of quantitative trading may seem trivial due to its similarity to regular trading, it should. However, there is a significant difference between a quantitative trader and a day trader: the use of algorithms to execute trades. At a very basic level and without delving too far into the technical nature of coding, an algorithm makes comparisons between various computer-stored values. In trading specifically, we are able to use this to our advantage by automating the aforementioned comparisons previously left to human judgment and emotion. This is useful, as little mental effort is required to complete these calculations, allowing for more time to be spent determining the conditions of interest and even developing more algorithms. Further, as a consequence of using algorithms, we increase the volume of trades completed in a day by an astronomical amount, as data about markets can sometimes come in as little as microsecond divisions.
It is also worth mentioning the importance of Artificial Intelligence (AI) and Machine Learning (ML) currently being used in quantitative trading. In a world so rich with data—and more importantly, big data—it would be unfavorable not to use the tools available to us when generating potential profits. AI and ML in quant trading is currently being employed to provide insight into certain alternative data events, portfolio management, and even fraud detection. While we will not cover any AI/ML techniques here, it is definitely important to keep in mind.
Another major aspect to the strategy behind quantitative trading is backtesting. The idea behind backtesting is simple; using data from the past to test an algorithm. To backtest, an algorithmic trading strategy must have already been written. Once the trading strategy is translated to code, it is given data from the past allowing for the code to make a trading decision as if the market conditions it was fed were current. When given enough data over a long period of time, the backtesting engine will either generate profits or loss. It is this distinction which provides quant traders with the information they need to decide whether or not to implement a strategy. When executing high volume trades like quant traders, it is incredibly important to test your algorithm so you don’t lose your money at the drop of a hat.
Not only is backtesting generally used to test the profitability of an algorithm, but it is also used to determine the correct parameters by which to run an algorithm at. For example, a Simple Moving Average (SMA) algorithm depends on parameters such as the ticker(s) of interest and the lookback window by which to calculate the average. It is very possible for an SMA strategy to generate next to no profits with one ticker and a surplus of profit in another, or even with different length lookback windows. Backtesting is the method by which to vary these parameters in order to find the optimal run conditions.
In addition to backtesting, there is also a testing strategy used by quant traders called forward performance testing. As the name suggests, forward performance testing uses real time data and paper accounts (theoretical money) to test an algorithm. To briefly compare the benefits of the two testing strategies: backtesting is useful in the sense that it is fast whereas forward performance testing is useful because it uses real-time data. In actuality, it can never hurt to use both testing schemas, as more confirmation that the algorithm will run as intended can never hurt. For a more detailed comparison of the two testing methods look here.
The software that MachineTrader.io is built upon allows for both backtesting and forward performance testing. Specifically, the paper account feature itself is an implementation of forward performance testing. We have set up the system infrastructure so that when a paper account is selected, by default, you are conducting forward performance testing. From the node red terminal this information is then reported to the front end MachineTrader website for more elegant reporting. The MachineTrader infrastructure also allows for backtesting using the correct Polygon.io date range formatting. By testing an algorithm in a paper account using a date range from the past dating back to as far as 2004, we are able to successfully backtest the algorithm. Further, our chief quant, Jerzy Pawlowski, is in the process of developing an engine specific to backtesting, automating the process even more.
With all of this information in mind, the process of quantitative trading is just a matter of executing these steps: market research, identifying a mathematical model, building an algorithm, testing (and testing again) the algorithm, and finally running the algorithm in real time. While it may seem simple and a relatively fast process, both the research step and the testing step can take an exhaustive amount of time if guaranteed profits are desired. However, in the case where profits are reaped, the benefits of algorithmic trading far outweigh the costs. For instance, once an algorithm is proven to be profitable, the scalability is incredibly high. For a lot of major financial institutions, it is this factor that makes quantitative trading such a lucrative aspect of their portfolio.
Following the current trend, quantitative trading will continue to garner attention as long as the practice still guarantees consistent returns on investment. While many institutions have already implemented certain features such as AI/ML and alternative data, as the digital world continues to advance, there will be a greater and greater emphasis and dependence on these currently underappreciated tools in quantitative trading. The future of financial technology lies in quantitative trading.