Artificial Intelligence for Investors: An Introduction

Meet your financial advisor powered by artificial intelligence:

Artificial intelligence is a term used frequently but defined rarely. In our introduction to artificial intelligence, we’ll use the term interchangeably with the term “machine learning.”

At its core, artificial intelligence is simply an advanced form of pattern recognition that can be used to make predictions.

As a result, a good way to start any AI project is to start by thinking of something you’d like to predict.

In artificial intelligence terms, the thing that you are trying to predict is known as the “dependent variable.”

In investing, a common dependent variable is the price of an asset, or its future return based on its current price.

After defining the dependent variable as precisely as we can, the next step is to think of data points that may influence or predict the dependent variable.

In investing, recent prices and volume, as well as a wide array of other metrics related to the underlying asset, can be used for predictive purposes.

These variables that are designed to predict the dependent variable are known as the “independent variables”.

This is where the real secret to artificial intelligence lies: finding the most meaningful independent variables.

There are many predictive algorithms that can be customized and used, but a simple algorithm with great data will almost always outperform a highly advanced, complex algorithm that is fed bad data. In sum, garbage in, garbage out.

Once the independent and dependent variables have been defined, they should be combined into a single table — essentially a spreadsheet, in which all the independent and dependent variables are columns and the rows represent their relationship to each other.

Before feeding this data to a predictive algorithm, a portion of the data should be cast aside. This data, known as the “holdout data”, can be used to evaluate the accuracy of any artificial intelligence model.

The rest of the data, known as the training data, can be fed to an artificial intelligence algorithm customized to the data set at hand. The algorithm can learn the pattern in the data, so that when fed the same independent variables again, it can generate a prediction.

The quality of these predictions can be can be evaluated against the holdout set.

As a general rule of thumb, the more closely the predictions align with the actual dependent variable in the holdout set, the better the artificial intelligence is.

SixJupiter has built an artificial intelligence system for predicting stock prices one year out. To subscribe for signals based on SixJupiter’s AI, visit

Add comment