Quantitative Trading Strategies – Backtesting Accurately

Best Quantitative Trading Strategies

Quantitative Trading Strategies

Once you have identified and mapped out the quantitative trading strategies you want to trade with, it is time to backtest it. This can be done by obtaining data that is historical in nature and carrying out some tests. The purpose of this exercise is to find out whether your quantitative trading strategies which you have identified will be profitable or not when applied in the “real world”.

One must however bear in mind that Backtesting is not a sure shot guarantee to success. There are many reasons for it. Quantitative trading deals with several biases that may or may not work when applied as an investment strategy. As you backtest your strategy, you must spend enough time on it and carefully consider each bias. In this article or book, we will attempt to give you a short overview of some of these biases.

The most common bias you will come across is the look-ahead bias, the optimization bias (also referred to as data-snooping bias) and the survivorship bias. The other aspect of back-testing that one must be wary of is the cleanliness of the available historical data. The data has an important role to play in back-testing because the quality of data determines the transaction cost.

For those who are just starting out with quant trading strategies, its best to look out for data sets that are freely available on platforms such as Yahoo Finance, but instead of focusing on where you can find data that is best suited for back-testing, let us tell you what you should look for in the data so as it make it relevant to your back-testing method.

Best Quantitative Trading Strategies Testing Techniques

The following are the main concerns that arise when historical data is being dealt with:

Accuracy – This is the most important part. One has to determine whether or not the data being used contains gross errors. One way to deal with this is use spike filters that can easily point out mistakes in the data. But it is highly recommended to source the same data from at least two data providers and compare them against each other to spot errors.

Survivorship bias – This is often a problem that exists in data that is available freely or comes cheap. If the data in question has a survivorship bias, it means that it contains assets that are no longer relevant. For instance, if you are dealing with equities, it could mean that there are stocks in your data that have long since been delisted or have gone bankrupt.

Corporate actions – Often, a newbie on the quantitative trading gets caught on the wrong foot because of a corporate action. This implies that logistical activities of the company are accounted for in the real calculation of returns. For instance, a stock split should not be accounted for in the true returns. Where there are such actions carried out by the company, a process called a back-adjustment is necessary to keep your results relevant.

Once you are sure about your data, you will need to backtest it on a software platform. Most people these days prefer to backtest the data on software platforms that are dedicated to the purpose of back testing their quantitative trading strategies. However one can also use a numerical platform or a programming language for this purpose. The true purpose of a backtest is to find out how good its performance is by industry standards.

The metrics for the same are the Sharpe ratio and the maximum drawdown. Sharpe ratio can be obtained when the average of the returns in excess are divided by the standard deviation of the same. Maximum drawdown, on the other hand, is the largest drop from peak to trough in curve of account equity. This is usually over an annual or bi-annual time frame.

Once your quantitative trading strategies are free of all biases to the extent possible, and has a good Sharpe and minimised drawdown, you are ready to build an execution strategy, the next step in quantitative trading.