Underfitting

Underfitting, also known as underlearning, is a problem that can occur during the training process of an algorithmic trading system. It refers to the situation where the model does not fit the historical data sufficiently, and therefore, cannot predict future results adequately.

chart trading
chart trading

Underfitting can occur for various reasons, such as insufficient data, lack of relevant features in the model, or the use of a model that is too simple to capture the complexity of the data. When an algorithmic trading system suffers from underfitting, it tends to perform poorly on both the training and validation sets.

To avoid underfitting, it's essential to ensure that the algorithmic trading system is correctly specified and trained with sufficient historical data. It's also crucial to consider the inclusion of relevant and pertinent features in the model and ensure that the model is not overly simplistic.

chart trading
chart trading

If underfitting occurs, different strategies can be tried to improve the performance of the algorithmic trading system, such as increasing the size of the dataset, adding relevant features, or increasing the complexity of the model. However, it's important to be careful not to overfit the model, which can lead to the opposite problem, overfitting.


You might be interested in