By Justin Kuepper
Now that you have a trading system designed and coded, it is time to test it to make sure that your coding is free of logical and technical errors. We will also look at something known as optimization - a feature in some trading programs that allows you to fine tune your trading rules to fit the stocks that you plan on trading.
Testing Your Trading System
The vast majority of trading applications that support programming languages also support testing tools. These tools are divided into two categories:
Technical testing tools search for technical errors in your code. For example, if you forget to add a semicolon after a statement, the technical testing tool will notify you that your statement is invalid.
The location of the technical testing tool depends on the trading application being used. MetaTrader displays an error or flawed results when you try to compile your code, while trading applications like Tradecision have a "code check" utility built into the interface that lets you check your code for errors before applying it.
Logical testing tools search for logical errors in your code. For example, if you happened to use a "greater than" sign instead of a "less than" sign (which is not a technical error), a logical testing tool will show you that your results don't make sense.
The most popular logical testing tool is the backtesting tool. This tool allows you to take past data and apply your trading system to that data. This gives you an idea of the following:
- Whether your trading system is a profitable one
- What conditions prove to be most profitable
- Where any errors in your rules might exist
Troubleshooting Your Trading System
As with any other type of programming, troubleshooting can be a tedious and difficult task. Finding errors in your code requires systematically sorting through your code to identify syntactical errors that, although often minor, can bring your program to a halt.
Here are some common errors to look for:
- Missing semicolons after statements - These have to be after every statement.
- Undefined variables - Remember that you have to declare them before you use them!
- Spelling mistakes - If any names or functions are spelled incorrectly, the trading application will return an error (see example below).
- Incorrect usage of (=) - Remember that "=" assigns one value to another value, while "==" means "equal to".
- Incorrect usage of built-in functions - Consult your trading application's documentation or application programming interface (API) to make sure that you are using the correct syntax.
Here we can see that Tradecision gives us the location (line and column) of the error, a description of the error and the type of error (in this case, it is syntactical). If we look at the expression, we can see that in column 8 "xrossBelow" is not a valid function. If we replace the "x" (which is in column 8) with a "c", then we will have valid code.
If we look at MetaTrader, we can see that the errors come up when we try to compile the program:
Here we can see that in the description it says the "BuyNow" variable wasn't defined. Double clicking on this error message will bring us to the specific location of the error in the code.
As you can see, most trading applications give you an easy way to locate technical errors and fix them. Fixing the errors simply involves systematically going through each error message and then recompiling the code and/or applying the trading system to your charts.
Optimizing Your Trading System
Some trading applications let you select variables to be optimized. Tradecision, for example, lets you easily select a variable and replace it with code that will attempt optimization. Optimization itself is simply a process that finds the optimal value for a particular trading system element based on past results and performance. Note that over-optimization results in trading systems that are unable to adapt to market conditions; therefore, it is important to only optimize a few important variables, not every variable!
Here is what the optimization feature looks like in Tradecision:
You can see that we declared two new variables and set them equal to "#". The "#" simply means that the trading program will replace this with the optimal number. Next, you can see that we used the new variables within our trading strategy. Finally, we set a range for the numbers (so that the program will not search to infinity).
Some other trading programs have features that operate in a similar way, allowing you to replace the numerical value with a "#" and telling the trading application to optimize it.
By now you should have developed a working trading system in which you can have confidence. In the next part of this series, you will learn how to apply your trading system to charts and how to use it to make trading decisions!
TradingAutomate your trades by translating your strategy into a language your computer can implement in this hands-free approach to investing.
ETFs & Mutual FundsDiscover three ways investors can use tracking error to measure performance for a mutual fund or ETF, whether indexed or actively managed.
TradingLearn to add structure to your trading methods with these six important steps.
TradingLearn more about Code Climate and its suite of static program analysis tools used by software developers to produce and maintain high-quality software code.
Personal FinanceIf tax rules and regulations are Greek to you, read on to learn how to decipher them.
TradingMost brokers will provide you with trade records, but it's also important to keep track on your own.
TradingBy blending good analysis with effective implementation, you can dramatically improve your profits in this market.
TradingThese steps will make you a more disciplined, smarter and, ultimately, wealthier trader.
TradingIf live trading results differ from testing, changes to the plan may be needed. Common problems include: trader error, fills and slippage, technical problems and unique trading conditions such ...
TradingAn important part of a trading plan is testing to determine what you can expect of its performance. Backtesting and forward performance testing will help you predict if your plan will succeed ...