Back Test


Testimonials
I am impressed by StrataSearch's capabilities - especially the tools provided to minimize curve fitting.

- N.M., E-mail

Read More Testimonials!

 

A back test is the evaluation of a trading strategy over a historical period of time. When the back test is complete, the trader is provided with performance statistics that help identify the performance of the strategy over that period. This information then allows the trader to extrapolate how the strategy will perform into the future.

There are many performance statistics that can be evaluated in a back test. The Average Annual Return may be one of the most common statistics, but traders also look at maximum drawdowns, holding periods, standard deviations, consecutive wins and losses, and many other statistics. There are also a wide variety of complex formulas used to evaluate a back test, such as K-Ratio, Z-Ratio, Sharpe Ratio, Return Retracement Ratio, and Monte Carlo Simulations. Each of these values gives the trader one additional view of the performance of their strategy over the historical period of time, and allows them to identify its performance to a very fine detail.

A back test is certainly helpful, and some would say a required step in any trading strategy evaluation, but it is not perfect. A condition called overfitting (or curve-fitting) can sometimes be present within the trading strategy. In such a cases, the back test may provide extremely positive performance numbers, but in fact the strategy is so perfectly optimized for the historical period that it could never perform as well against anything but identical data. This can be a trap for traders who believe the results of their back test, only to find that their system fails moving forward.
Back Test
Back Test Performance Report

A back test is certainly helpful, and some would say a required step in any trading strategy evaluation, but it is not perfect. A condition called overfitting (or curve-fitting) can sometimes be present within the trading strategy. In such a cases, the back test may provide extremely positive performance numbers, but in fact the strategy is so perfectly optimized for the historical period that it could never perform as well against anything but identical data. This can be a trap for traders who believe the results of their back test, only to find that their system fails moving forward.

Although there are many software packages that can run a back test, there are far fewer that can protect against overfitting. Programs like StrataSearch, for example, allow the user to perform many tests to identify how robust a strategy is. As an example, a back test can be run against alternate time periods and even alternate groups of symbols. Such out-of-sample testing can identify whether the successful numbers can be seen on a wide variety of trading scenarios. Even further, StrataSearch allows you to run a back test using shifted parameters, where you might try a 13-day and 15-day Relative Strength Index alongside your 14-day Relative Strength Index.

Back Test 2
Back Test Account Equity Over Time

In addition to a back test, many software packages also allow the use of a forward test, where performance is evaluated on the data immediately following the data used to create the strategy. Such testing can provide an additional level of evaluation apart from the main historical period.

The availability of powerful computers has made it possible for the individual trader to build trading systems and back test their strategies quite easily. And while there is a great deal to learn regarding which performance statistics in the back test identify a winning trading system, the back test may be the most important tool for traders interested in system trading.