In my experience, backtesting a trading strategy is an essential step before deploying it in the market.
It allows me to assess how the strategy would have performed in the past, giving me a clearer idea of its potential success in the future.
Manual backtesting is great, but automated backtesting speeds up the process considerably.
There are three ways to automate backtesting of a trading strategy: learn a programming language, use pseudocode software or use no code software. Each one has pros and cons that must be considered when choosing an automated backtesting method.
So in this article, I’ll explain each of the options and how to figure out which one is the best one for you.
Keep in mind that you don’t necessarily need to learn programming to do an automated backtest, and this can be done in any trading market where sufficient historical data is available.
I’ll also provide links to the best resources that I know of, based on my experience and research.
Let’s get into it…
Key Takeaways
- Automated backtesting can be done even if a trader doesn’t want to learn how to code.
- Not all trading strategies can be fully automated. But in those cases, semi-automated backtesting can be extremely helpful.
- There are many free and paid tools that can help traders of all skill levels get into automation.
Why Backtesting a Trading Strategy is Important
I’ve written many articles and tutorials on backtesting.
You can find them here.
But if you’re new to the idea of backtesting, I’ll give you the elevator version of why it’s so important.
By utilizing backtesting software, I’m able to simulate years of trading within minutes, allowing me to test multiple strategies against a wide range of market conditions.
The software does the heavy lifting, executing my strategy across the selected time period and accumulating the results for my analysis.
What I look for in the backtest results is not just overall performance, but also how the strategy behaves during different market phases.
Whether the market is trending, ranging, or experiencing high volatility, I need to know that my strategy can handle these conditions.
The insights gained are invaluable.
They either prove that the strategy has a legit edge, or I go back to the drawing board to make improvements.
Why Automated Backtesting?
The bottom line is that automated backtesting helps me speed up the process of creating and optimizing profitable trading strategies.
I first started backtesting with a simple chart and a spreadsheet.
That was fine in the beginning because there weren’t the tools that are available now.
But it would take foreeeever to backtest just one market, even on the daily chart.
When I can automate backtesting, I can speed up that process dramatically and test multiple markets in a fraction of the time.
What many traders also miss is that they can do semi-automated backtesting.
Since not all trading strategies can be 100% automated, semi-automated backtesting allows me to automate just the entry, for example.
Then I can use discretion to exit the trades.
Any aspect of a trading strategy that can be semi-automated, should be, because it saves a ton of time.
Benefits of Automated Backtesting
Automating the backtesting process offers efficiency and the ability to test numerous strategies across different timeframes and market conditions very quickly.
Many software packages will also provide detailed analytics on each backtest.
This allows me to focus on coming up with ideas and optimizing, instead of figuring out how to get these metrics in a spreadsheet.
The tools available nowadays can make this easy to do.
Limitations of Automated Backtesting
Despite its many benefits, there are limitations to automated backtesting.
One primary challenge is ensuring the historical data used is complete and accurate.
Any gaps or inaccuracies can significantly alter the test results.
There’s also the risk of overfitting a strategy to past data, making it less adaptable to future live market conditions.
Another point to consider is simulation limitations, as models may not account for factors like market impact, liquidity, or real-time transaction fees.
Finally, not all trading strategy criteria can be completely programmed into a computer.
The First Step Before Backtesting
Before you start backtesting with one of the solutions below, be sure that you have a trading strategy to test.
This means having a complete trading plan.
You need to have specific parameters for your entries, exits, risk and trade management.
Once you have a strategy to test, now it’s time to pick a platform to help you do an automated backtest.
Automated Backtesting Software Tools
Before picking a robust backtesting framework, it’s essential to understand your options.
So here are the 3 types of tools that are available, along with some resources in each category.
New solutions are constantly popping up, so use this list as a starting point in your exploration.
I’ll update this list as frequently as possible, but it’s up to you to find the best solution for your situation.
Learn a Trading-Friendly Programming Language
To effectively develop and customize a trading algorithm, you must be fluent in a programming language commonly used by traders.
I haven’t had the patience to learn a programming language myself, so I’m not an expert in this area.
But I have done quite a bit of research on the topic and here’s what you need to know.
Benefits of Programming
- There are highly mature languages that have been proven to work for trading
- Many free tools available
- The ultimate level of customization capabilities
- You can build reusable blocks of code that you can use for future backtests
- There is a large library of open source code and tutorials that you can use to get started quickly
- Can be used in any market where historical data is available
Downsides of Programming
- It can take time to learn how to code at a level that will allow you to create meaningful backtests
- Creating simple things like performance graphs usually takes extra time to learn
- One misplaced semi-colon can break the entire program and take extra time to find and fix
There are 3 programming languages that are frequently used by professional traders:
- Python
- R
- C#
I’ll give you a quick summary of each one and where you can go to get more information.
Python Programming Language
I personally know a couple of traders who build their trading strategies with Python.
It emphasizes code readability and allows programmers to express concepts in fewer lines of code compared to other languages.
Python supports multiple programming paradigms, including procedural, object-oriented, and functional programming.
It has a vast standard library and a thriving ecosystem of third-party packages.
Python’s interpreted nature facilitates rapid development and debugging, making it popular among both beginners and seasoned developers.
Here are resources that will get you started with programming in Python:
- Python beginner’s guide
- How to use the Python backtesting framework
R Programming Language
R is a programming language and environment that is primarily used for statistical computing and graphics.
Perfect for trading.
It provides a wide variety of statistical and graphical techniques, including linear and nonlinear modeling, time-series analysis, and clustering.
R is open-source and highly extensible, with thousands of packages available for specialized tasks.
It is widely used in academia, research, and industry for data analysis, visualization, and machine learning applications.
R’s syntax is designed to make data manipulation and analysis intuitive and efficient, making it a popular choice among statisticians and data scientists.
Here are some resources that will get you started with programming in R:
C# Programming Language
C# is a versatile, object-oriented programming language developed by Microsoft as part of its .NET framework, designed to be simple, modern, and type-safe.
It’s widely used for developing desktop applications, web services, and enterprise software, and it has extensive libraries that facilitate tasks ranging from database interaction to web development.
Traders use C# to program automated backtests by writing algorithms within trading platforms like WealthLab, or QuantConnect, which support C# natively.
They utilize the language’s robust features to define trading signals, manage orders, and analyze historical data for performance evaluation.
The backtesting environment provided by these platforms allows for the simulation of trading strategies over past market data to assess their efficacy before live deployment.
Here are some R resources that will get you started with programming in C#:
Now if programming isn’t your cup of tea, then luckily there are other options out there.
Use Pseudocode Software
Pseudocode software serves as an intermediary solution between programming and no code software.
It’s meant to be a little easier than traditional programming, but still give you more control and customization than the no code route.
There are many built-in shortcuts that are included in each platform, reducing the amount of code needed for each application.
I’m personally not a fan of these solutions either, because they are still too much like programming for my taste.
However, this is useful to some traders who want to automate their backtesting, without learning all the nuances of a true programming language.
So if you don’t want to become a full-on programmer, then this might be a happy medium.
Benefits of Pseudocode Software
- Easier to learn than programming
- Usually more built in tools for things like graphs and statistical analysis
- Less code needed when compared to traditional programming
Downsides of Pseudocode Software
- Platform dependent, so your pseudocode will only work on one platform
- You may not be able to do everything you can do with a regular programming language, since there isn’t the same level of customization available
I’ve personally used 4 solutions:
- Pine Script
- MQL
- EasyLanguage
- AFL
Here’s a summary of each of them and where to get more information.
The right one for you will depend on the platform you use and the market you trade.
MetaTrader / MQL
MQL, or MetaQuotes Language, is a proprietary scripting language for developing trading robots, technical indicators, scripts, and function libraries for use with the MetaTrader software.
MQL4 is used with MetaTrader 4, while MQL5, its successor, offers more advanced capabilities and is used with MetaTrader 5.
Both versions of MQL are specifically designed for automated trading, strategy testing, and financial market analysis.
This is a very popular language that some people might consider a full-on programming language.
In my opinion, it falls more into the category of pseudocode because you don’t need to include all of the elements that are required for a normal computer program.
It’s fairly complex however, so it does take some time to learn.
Here are some resources that will get you started with MQL:
TradingView / Pine Script
Pine Script is TradingView’s platform-specific scripting language used for creating custom technical analysis indicators and strategies directly within its charting platform.
It’s designed to be user friendly, allowing traders with minimal programming experience to design, test, and share their trading ideas and systems.
Pine Script enables the customization of charts and the creation of alerts based on specific trading conditions.
I’ve created a few custom indicators with Pine Script and it is easier than MQL.
Here are some resources that will get you started with Pine Script:
TradeStation / EasyLanguage
EasyLanguage is a proprietary programming language used in the TradeStation platform for developing and backtesting trading strategies.
It’s designed to be user-friendly and accessible to traders without extensive programming experience.
EasyLanguage allows traders to create custom indicators, strategies, and trading systems using a simplified syntax and built-in functions.
Here are some resources for learning TradeStation’s EasyLanguage:
Amibroker / AmiBroker Formula Language (AFL)
AmiBroker Formula Language (AFL) is a scripting language used in the AmiBroker trading platform to create custom indicators, trading strategies, and analytical tools.
This solution is primarily for stock traders.
AFL allows users to define and manage charts, scan for opportunities, and backtest trading systems with speed and precision.
It’s designed to be easy to understand for those familiar with technical analysis and trading concepts, while also being powerful enough to implement complex algorithms.
Here are some resources for learning Amibroker’s AFL:
Use No Code Software
No code backtesting software allows traders to test their trading strategies using a visual interface without writing any code, typically by selecting indicators, setting parameters, and defining rules through dropdown menus and form fields.
These platforms often provide a user friendly drag-and-drop or point-and-click environment to facilitate the creation and historical evaluation of trading strategies.
They aim to make backtesting accessible to those without programming skills, while still offering robust analysis and performance metrics.
Benefits of No Code Software
- Extremely easy to use
- Can start creating meaningful backtest very quickly
- Some platforms have a wide range of trading strategy inputs to create many different strategies
Downsides of No Code Software
- Lack the same level of customization that programming or pseudo coding provides
- Some platforms are very limited in what they can do
I’ll share my experience with 2 platforms, but do your own research because new ones are popping up all the time.
NakedMarkets
This is my favorite no code platform.
It has a ton of flexibility that I haven’t seen before in a no code platform.
I can create fairly complex trading strategy logic with its intuitive drag and drop interface.
You can backtest Forex, stocks, indexes and more.
The detailed analytics suite also provides a ton of data on a backtest that you don’t get in other platforms.
Everything can be backtested in the platform, so there’s no need to export an EA or plugin to another platform for backtesting.
They are also working on an export feature that will export automated trading strategies to MetaTrader.
Check out the videos and tutorials below to get a feel for what the program can do.
You can also get a discount and a special offer on NakedMarkets here.
Here are some resources for learning NakedMarkets:
EA Builder
EA Builder is an online platform that allows traders to build MetaTrader 4 and 5 Expert Advisors (EAs), and TradeStation strategies, without coding.
The benefit is that the software will export the strategy that you build on the website to a file that you can import into MetaTrader or TradeStation.
I found that this helps a lot with very simple trading strategies.
But once you try to build more complex ones, the platform simply doesn’t have the capacity to do that.
So if you’re looking for something to build simple strategies and backtest them in MetaTrader or TradeStation, then this could be for you.
However, you should try their free demo first to see if this is a good fit for what you want to do.
Here are some resources for learning EA Builder:
Evaluating Backtesting Results
Automated backtesting results have to be evaluated a little differently than manual backtesting results.
The first thing I look for are potential errors in the logic of the trading strategy.
Since the trades are being executed automatically, I cannot see the trades in action.
So I have to double check that the trades look correct.
I do this by looking at the trades list and I spot check a few of the executions manually.
If that checks out, then here’s what I look at:
- Profitability: I not only look at the overall profit of the strategy, I also look at how consistent the strategies is on a yearly basis. A strategy can be profitable, but all of that profit could have come from just one month. I like to see steady growth over multiple years.
- Win / Loss Ratio: I also take note of the win-loss ratio which indicates the proportion of winning trades to losing ones. A low ratio is not necessarily bad if the winners are much bigger than the losers, but I want to see if there is any room for improvement.
- Drawdown: Understanding the drawdown of a strategy is vital. It involves assessing the largest peak-to-trough decline in my account balance. I calculate the maximum drawdown and if it’s too large, it indicates potential high risk, and I should probably adjust my risk management approach.
- Monte Carlo Simulation: I run a simulation over at least 100 simulations to get a good idea of the maximum potential drawdown.
- MFE/MAE: These metrics show how much the trades went in my favor or went against me, while the trades were open. This can give me valuable clues as to how to improve the strategy.
- Risk of Ruin: This shows my chances of blowing out an account with a strategy. I want the risk of a huge loss to be fairly low. Other traders can tolerate a more risky strategy.
- In and Out of Sample Data: It helps to test a strategy with only part of the historical data, then test the optimized strategy with the rest of the data. If the strategy works well with both data sets, then the strategy is more robust and probably more reliable.
- Time of Day or Day of the Week Analysis: I look at how profitable certain days are, or how profitable certain hours of the day are, if I’m testing a day trading strategy. This will show me if I might want to stop trading certain days of the week or hours of the day.
The reality of backtesting is that most trading strategies won’t be profitable enough to trade with real money.
But by reviewing the metrics above, I can potentially spot ways to turn a failed strategy into a profitable one.
I can also potentially make a profitable strategy even more profitable.
Final Thoughts on How to Automate Backtesting
Automated backtesting can take your backtesting process to the next level by allowing you to test a ton of ideas in a very short amount of time.
In addition, you can leverage semi-automated backtesting to test specific aspects of a trading strategy, if the entire strategy cannot be automated.
The technology nowadays is quite good, so I feel that all traders should at least do some automated backtesting because it’s so easy and fast.
So if automated backtesting appeals to you, take the time to learn your preferred platform.
It’s one of the greatest skills that you can have as a trader.
The post How to Automate Backtesting of a Trading Strategy appeared first on Trading Heroes.
