ChatGPT has been on everyone’s mind lately. People are discussing how it can improve productivity or help you solve programming bugs. But can ChatGPT be used for backtesting?
ChatGPT can be helpful in trading as a tool for analyzing market data, generating trading ideas, and backtesting trading strategies. But it can be tricky to get it to understand what we want (and need). So, how can we implement it into our trading process?
In this article, we will look at what ChatGPT is and how to use it for trading and backtesting a trading strategy generated by AI (artificial intelligence). Furthermore, we find out if ChatGPT can help in trading and how you can backtest your trading strategy.
What is ChatGPT?
ChatGPT is an AI system developed by OpenAI launched in November 2022. The best part about it is that you can interact with it conversationally. The dialog format allows ChatGPT to answer follow-up questions, admit mistakes, challenge incorrect premises, and reject inappropriate requests. It’s some sort of human behavior.
It attracted a million users only five days after launching, making it the fastest product to reach that amount in history, surpassing companies like Instagram and TikTok. In January 2023, only two months after its launch, ChatGPT had 100 million active users, according to an analysis by UBS. For comparison, it took nine months for TikTok to reach 100 million.
In March 2023, they launched ChatGPT 4, an upgraded, better version of the model available for a subscription of $20 per month. However, there’s still a free version of ChatGPT 3 available for anyone.
How to use ChatGPT for trading
If you expect ChatGPT to generate a super profitable trading strategy just by simply asking it, let us tell you that you are using it wrong, which will be a huge disappointment. As you wouldn’t expect a new intern to know how to do their job, you can’t expect an AI machine to know how to do its job.
Today we will ask it to brainstorm different trading strategies and backtest one of them.
Firstly, to unlock ChatGPT to its full potential, we first need to train it to do what we want. What do we mean by that? Tell it what you will use it for and what you want. Here’s a screenshot:
We know this may seem silly, but now the AI machine knows what we expect to get from it.
Next, we could give it some basic trading ideas and see what it comes up with:
As you can see, ChatGPT can sometimes give incorrect answers. The mean reversion and trend following strategy are virtually the same. We can ask it to expand and provide other mean reversion ideas in this case:
The answers aren’t always very detailed, but you can always ask it to dig deeper. For now, we will just backtest one of the mean reversion strategies.
Backtesting a trading strategy generated by Chat GPT
We will ask it to backtest the Bollinger band mean reversion strategy. We won’t upload the answer because it’s too long, but we wrote the following prompt:
Let’s go with the Bollinger Band strategy. Write a python program to backtest the strategy using pandas, numpy, yfinance, and matplotlib.
Then we copied the code and ran it on Python without changing a thing.
The strategy that ChatGPT backtested is the following:
- It sets the Bollinger Bands parameters to a period of 20 and a deviation factor of 2.
- It buys when the price closes below the lower band and sells it when it closes above the upper band.
For the backtest, we have used SPY since its inception. The data is adjusted for dividends and splits. Here is the equity curve ChatGPT returned:
That is a very decent equity curve! (What is a good equity curve?)
ChatGPT ends the backtest with this graph, but we calculate some trading performance metrics and statistics on our own: The CAGR of the strategy is 3.20%, it made 326 trades, and the time spent in the market is 45.63%.
Of course, we can’t expect ChatGPT to hand us a killer strategy, but it provides a great starting point to build our own models and strategies.
How to backtest a trading strategy using ChatGPT – conclusion
ChatGPT is a truly revolutionary program that will surely mark a “before and after” in the history of AI. We already see ways it can help with our day-to-day tasks, such as programming, writing, or researching.
But for now, it’s constrained for trading. It’s important to note that ChatGPT is not a trading platform, and it does not provide financial advice or make investment decisions.
The most helpful way we can implement it in our day-to-day tasks as traders are by brainstorming trading ideas or generating backtest in Python (or the coding platform of your choice). Nevertheless, it won’t take long before AI tools like this become a central part of quantitative models.
List of investment strategies
We have written over 1200 articles on this blog since we started in 2012. Many articles contain specific investment rules that can be backtested for profitability and performance metrics.
The trading rules are compiled into a package where you can purchase all of them (recommended) or just a few of your choice. We have hundreds of investment ideas in the compilation.
The strategies are taken from our landing page of investment strategies.
For a list of the strategies we have made, please click on the green banner:
These strategies must not be misunderstood for the premium strategies that we charge a fee for:
Related reading: ChatGPT Trading Strategies (How To Use AI To Make Strategies)