The inspiration for this project stems from the interest to continue learning after completing in June 2024 a 6-month Professional Certification at Imperial Business School on Machine Learning and Artificial Intelligence.
I continue experimenting with neural networks on this project. I am privileged to be guided by Ali Muhammad, who lectured me during my certification at Imperial.
In this project we use different neural network approaches to estimate gap risk on the price of financial assets, and each approach is held in a subdirectory in this repo. The projects in this repo may slightly deviate from this objective as I explore and research associated predictions that help me build towards the end goal.
The project is work in progress. This proof of concept (PoC) prioritizes demonstrating feasibility. Code quality and refinement will be improved in future updates. Future iterations will focus on refining the code and implementing best practices.
You can follow the journey in this blog.
CNN-share-price-prediction: Convolutional Neural Network to predict next day share price from a stock price time series, with Bayesian hyperparameter optimization
The project includes a LeNet-5-based design, which is a Convolutional Neural Network (CNN) with both convolutional and linear (fully connected) layers, to predict next day share price from a stock price time series. Stock financial time series are encoded in Gramian Angular Field images and used as inputs.
If you find this helpful you can buy me a coffee :)
