08/08/2021 In this repository, my goal is to expand upon the knowledge have gained in the domains of finance, software engineering, data science, and machine learning to create a more professional and usable product, yielding informative results and applicable systems.
08/26/2021 So far, there isn't much to see here. We have utilities to load and clean data, as well as utilities to build and train a basic neural network. Our baseline model was able to predict the direction of price change with an accuracy of ~53% (Which can be seen in this notebook github.com/voynow/finance/blob/main/preliminary_modeling.ipynb). This is not great. Typically here we could iterate on more complex models and try to squeeze out a few more percentages, but that would only be a short term solution. Instead of continuing the modeling phase, let's take a step back and consider possible next steps with our long term goal in mind. If we want to build professional scale software, we can't rely on the Yahoo Finance API for all of our data needs (especially if we want to deploy software in production). We need to build our own database, which means that I need to learn SQL.
9/11/2021 Building the database was instructive. I have studied SQL theory in the past, but have not had the opportunity to proceed with a python implementation until now. With our databse spun up, we can start thinking about potential applications. It is important to note that I decided to ditch the deep learning route (for now). Reflecting on the project goals, it seems that deep learning and price prediction doesn't fit organically into this repository.
9/19/2021 At the moment, the development of this repository will be on pause. The goal of building an end to end product was ambitious, but the goal of learning was sufficiently achieved. As previously mentioned, this was my first time working with relational databases in python. By no means am I an SQL expert, although I am now confident enough in my ability in order to use it properly/effectively in future projects.