Oh boy, have we bitten off more than we can chew! Is this AI/ML bitcoin trading bot project going to stretch our mental resources and coding skill set to the breaking point? We had a group meeting where we explored and talked through some of the possible avenues for tackling the project and settled broadly upon building a simple trading bot that uses a small set of simple financial indicators to make and publish trades. From there, we will integrate the neural network and refine our trading bot to make its trades more sophisticated. Easier said than done.
What we quickly realized is that the more we’ve learned and come to know, the more we learn about what we don’t know – the grand vastness of what we don’t know. This is trivially true of any real worthwhile endeavor in a broad and complex field, but this seems especially true of machine learning. It’s time for me to don my big points and drink from the fire hydrant of knowledge. YouTube Python tutorial after Python tutorial of how to connect with and talk with the Binance API , how to use TensorFlow, refine my NumPy and Pandas skillset, etc.
If anything, this is really learning about two vast fields, trading and machine learning. I have had relatively limited trading, but this is a vast field unto itself. Whilst learning how to build a simple trading bot hooked up to the Binance API, I have had to learn how to interpret financial indicators like relative strength index and make calculations on them, interpret them, and feed them into datasets to be further parsed and analyzed.
And this is before diving deep into the AI research. Without having taken linear algebra, some of the technical and logical fundamentals of tensors, matrices and manipulations of data containers will have to be somewhat blackboxed. What is also daunting is the intense review of college statistics, distribution curves, how to normalize data and the other pieces of data science that overlap with the study of machine learning.
Hopefully, this does not read like self-defeat. To the contrary, this is a thrilling if nerve wracking way to hit the ground running in a field I have deep interest in. As I mentioned in the previous blog post, this is a field that I would like to pursue in graduate education and in my future career. Already, this is proving to be more stimulating than the more mundane drudgery of adjusting button placement in JS/React.
