We need a reliable dataset to train our monkey on; preferrably highly granular, and extending pretty far back.
Luckily, my preferred brokerage Alpaca offers minutely candlestick bars going all the way back to
December 2013. Now we could just use any other pre-made dataset offered by anyone, but I want to showcase an
understanding of real data-science principles and work ethic here so I'm rolling my own.
The PriceData and DatabaseOperations submodules further describe the method behind data curation in greater
detail.
Once we have an agent trained on the historic data and tweak it to our liking we need to keep it updated with
current data as it comes in. The Alpaca historic data client only goes up to fifteen minutes ago, but also
they offer live bid/ask/amount transactions as new orders get placed on the books. This adds a layer of complexity,
as I'm not quite sure on the mechanism behind interpreting this kind of data. From what I gather, it's similar to
the one about predicting housing prices in a given area using linear regression.
Anyways, we need to be able to interpret this live market data as quickly as possible so we can make a decision while the information is still relevant.