Quantitative trading strategy refers to finding high probability and effective trading strategies in the market, by using many mathematical and statistical tools through computer data analysis, model building, back-test verification, transaction execution, optimization, and other processes. A rational, objective and automated trading process can be achieved without relying on human subjective judgments, therefore, quantitative trading strategy is often referred to as automated trading.
With the invention of integrated circuits and the development of computer science, people began to explore the possibility of applying the powerful data processing and computing power of computers to the financial trading market. Harry Max Markowitz, the Nobel Laureate in economics, is also known as the father of quantitative trading strategy. In his representative paper “Portfolio Selection”, he discussed the efficiency of asset allocation in a numerical way, and assisted two fund managers to execute the first computer arbitrage transaction in the financial market.
From 1970 to 1980, quantitative trading strategy began to emerge. The New York Stock Exchange adopted the Designated Order Turnaround (DOT) system, which greatly shortened the delay of investors’ orders and improved operational efficiency. Since 1990, algorithm systems have become more and more popular, and many hedge funds also invested in the arms of quantitative trading strategy. The dot-com bubble in 2000 proved the effectiveness and strength of quantitative strategy trading. When the market was still immersed in the last party, quantitative trading strategy helped investment institutions reduce their positions in high-risk dot-com stocks and successfully avoided the subsequent market collapse.
According to statistics, in 2010, more than 60% of the trading volume of the U.S. stock market came from high-frequency trading investors and market makers using quantitative strategies. After decades of development, automated trading robots have now accounted for half of the financial market.
The automatically executed quantitative trading strategy has the following advantages over the ordinary trades in which users make transactions by themselves:
There are many types of quantitative trading strategies. The following are some common quantitative strategies in the field of digital currency:
The quantitative strategy refers to dividing the assets into equal parts according to the set grid quantity, and pend orders at different grid prices within the set price range. When the market fluctuation crosses different grid prices, the program automatically buys and sells in batches, so as to earn profits from grid differentials.
The quantitative strategy is similar to the index fund, which combines different investment objects according to the selected proportion, sells assets with high positions and buys assets with low positions when the market price changes, and dynamically adjusts the positions to restore the initial proportion of each investment object, so as to obtain long-term stable returns.
Due to the capital rate in the perpetual contract market, when there is a difference between the futures price and the spot price, the futures and spot hedging can be carried out to earn the current price difference that decreases over time. For example, when the capital rate is positive, buying a certain value of spot and issuing a futures short order of equal value can offset the gains and losses caused by the rise and fall, and obtain the capital rate returns of the perpetual contract market.
Commodity Trading Advisor (CTA) Strategy:
The quantitative strategy refers to using a single or multiple technical indicators for market monitoring. When the collected transaction data meets the set conditions of the indicators, the transaction signal will be triggered, and the program will automatically make transactions.
Sell High Buy Low for Arbitrage
Most currencies can be traded on different platforms. Due to the differences in pricing methods, trading volume and market depth, sometimes the same currency has different quotations on different platforms. Sell high buy low for arbitrage refers to a behavior of buying on a lower price platform and selling on a higher price platform to earn a price difference. The opportunity to sell high buy low for arbitrage is fleeting, and multiple trading platforms need to be monitored in real time, so it is usually completed through the calculus of high-frequency trading.
Formulating a quantitative trading strategy usually includes the following steps:
Any quantitative strategy needs to have clear profit means and dimensions, such as earning spread, volatility, time value, arbitrage, and so on. The strategic idea can collect a large amount of market data for specific parameters for statistical analysis and model building.
After collecting enough data, we can start data exploration. At this stage, mathematical statistical tools will be used to do outlier screening, clustering, analysis of variance (ANOVA), regression analysis, or machine learning algorithms, etc., so as to find out the rules and formulas hidden in big data that can be used as trading strategies.
Data backtesting is a necessary process before any quantitative strategy is officially launched and operated. It can evaluate various data performances of quantitative strategy, including the winning rate, profit/loss ratio, performance curve, maximum fallback, invalid factor and so on. Good data backtesting can help quantitative strategy designers find potential problems as soon as possible, so as to optimize and iterate the subsequent model.
If the quantitative trading strategy does not pass through the practical experience of the trading market, it will eventually become a moot point. Some platforms provide paper trade, so that users can use MIMIK funds to record profits and losses according to the actual market conditions, and confirm whether the constructed quantitative strategy meets the expected stable profits.
Although quantitative strategy trading has brought many conveniences and advantages to its users, it is still necessary to pay attention to the possibility that other risk factors may cause quantitative trading strategies to fail. The stability of service providers is a very important link. In case of equipment failure or network interruption, it will not only cause the quantitative strategy program to fail to operate normally, but even cause risk exposure and property losses due to the inability to close positions. The source of quotation data is the same as the network hacker attack. The wrong quotation data will lead to the misjudgment of the program, and the loopholes and algorithm defects of the program code will be attacked by other participants in the market and cause losses.
Due to the increase in the number of quantitative strategies and the complexity of the model, there may be correlations and unexpected interactions between different strategies and different trading parameters. Regular update maintenance and backtesting inspection are necessary. In some transactions with large capital volume or high risk, the quantitative strategy is only used as the reference basis for the operators to open and close positions, rather than fully automatic operation. For such quantitative auxiliary transactions, there must be a perfect standard operating process and education and training to avoid the negligence of manual operation.
Before using the quantitative trading strategy, we must understand that the quantitative trading strategy is not a panacea for any market or quotation. If the effective parameter index in the traditional financial market is changed to the cryptocurrency market, it may fail. In addition, the historical backtesting results of any quantitative trading strategy cannot be used as a guarantee of future performance. For example, when a trading strategy attracts many investors because of its excellent performance, many traders in the market will rush to buy and sell at the same point, making the strategy that should have generated profits become unprofitable.
In addition, trading is not only a profound knowledge, but also an art. Some top professional traders do not completely rely on objective index data when performing entry and exit judgments, but also rely on abstract “stock feeling” sometimes. Although the rapid development of artificial intelligence has reached a level far beyond human beings in the fields of complete information games, such as chess, Japanese chess, and Go, it is still impossible to demonstrate the so-called “intuition” and “sixth sense” in the chaotic incomplete information trading market.
The performance of traders depends on their personal experience and ability, and quantitative strategy trading is no exception. Quantitative trading strategy written by developers without sufficient professional knowledge and experience is difficult to have good performance. The design of quantitative strategy involves many different fields, so we must have considerable professional knowledge of mathematics, statistics, finance, computers and so on, so as to develop excellent quantitative trading strategies.
Quantitative trading strategy does not necessarily have to use complex high-end algorithms. In fact, as long as there is a fixed transaction logic in any trading behavior, developers can write code to automate its execution process. The most common is the grid trading strategy, which is very suitable for automated procedures to replace manual operations because it mechanically hangs the purchase and sale orders back and forth.
Quantitative strategies are also suitable as an auxiliary reference for manual trading judgment. Modern financial markets are changing rapidly, and it is inappropriate to rely on one’s own efforts to digest a large amount of information to make investment decisions. Making good use of the huge information aggregation ability and statistical tools of computers can provide users with a broader vision to find better trading opportunities.
The emergence of quantitative trading has also contributed to the development of high-frequency trading.
High-frequency trading refers to the fact that the automatic program performs many trading operations in a very short time. According to the changes of the market, the high-frequency trading robot can even make the judgment of long and short conversion in one thousandth of a second, and carry out a series of pending orders and cancellations. That is, high-frequency trading, which maximizes the efficiency of capital use through a large number of instantaneous transactions by making the holding time tend to zero risk.
The purpose of high-frequency trading is to find fleeting trading opportunities and small profits that human beings cannot capture from the daily fluctuations of prices. Due to the rapid development of computer science, high-frequency trading is a highly demanding and competitive field, and there are many requirements for equipment upgrading and algorithm optimization. Even if the arbitrage program uses the same code, if the sampling rate of market information is different, or the performance of equipment is different, it may lead to different results that one person gains while the other loses. Generally speaking, the higher the sampling rate of market information and the faster the program execution speed, the more advantageous of high-frequency algorithms in the trading market.
High-frequency trading has accounted for a considerable proportion of the trading volume in the global financial market. It reduces the market spread and provides a lot of liquidity. However, the competition between different high-frequency trading procedures has also increased the fluctuation of market prices. High-frequency trading algorithms are generally complex and difficult to develop. Usually, only large financial institutions or market makers have such quantitative trading strategy tools.
With the development of the computer field and the innovation of financial derivatives, professional investment management teams and market makers have begun to adopt automated procedures for quantitative trading. Compared with the general transaction of traditional manual transaction, quantitative trading strategy has many advantages, such as abiding by discipline, fast execution, consistent logic, objective decision-making, non-stop throughout the year, easy performance verification, synchronous monitoring of a large number of trading markets, self-learning and so on.
However, the cross-domain knowledge necessary for the development of quantitative strategies and the increasingly fierce competition also make the threshold of quantitative trading strategy higher and higher, and the faults and defects on the equipment, network, code and model during its operation are also factors that must be considered.
Presently, quantitative trading has occupied a place in the global financial market. How to make the long-term asset curve grow steadily and avoid the performance being washed up and down like a roller coaster due to market fluctuations is the goal of most top quantitative strategies and teams. In addition to algorithm iteration and developing new markets, high frequency, high winning rate, low risk and arbitrage accumulation will become the future development trend of quantitative strategies.
Quantitative trading strategy is not the holy grail, and there is no guarantee of profit. Like traditional ordinary trading, it will encounter the risk of loss. Only after knowing the advantages and disadvantages can we control this tool well.