Backtesting is a powerful tool that can be used to help facilitate future decisions based off of past information. According to an article on Medium, “Backtesting is the process of applying a trading strategy, predictive model, or analytical model to historical data to evaluate its accuracy and performance.” Put simply, backtesting is an analyzation method that can be used to determine future decisions based off of past data. This process can be used in the stock market to allow users to make more educated decisions on when to buy or sell a stock.
The process of Backtesting is implemented into machine learning programs by allowing the program to analyze past data and make a prediction for the future based off of the information given. Machine learning is used in this process when the program decides whether or not the user should buy, sell, or hold on the stock depending on the information given to it. When implementing backtesting into machine learning, there are two approaches that can be taken; event-driven or vectorized backtesting.
The general trading strategy that backtesting programs follow require inputs from the user to help guide the decision. Generally, these programs require an amount of days to predict in advance, a price target, a closing price, and past information of the stock. The backtesting programs will follow these guidelines on making their stock judgements.
What makes backtesting with a program much better than backtesting manually is that a program has the ability to analyze a much larger amount of data at a much faster rate. What this means is that backtesting with a program gives the user a much faster judgement time on whether or not to buy the stock compared to a user who has to manually look over data. This extra time given can allow the backtesting program user to make quick decisions on whether stock trading instead of having to wait long amounts of time analyzing data themselves.
Source: https://machinelearningmastery.com/backtest-machine-learning-models-time-series-forecasting/