So this past week has been yet another eventful week for crypto enthusiasts, the weekend pump and dump had us (my self included) in high hopes. Never the less there were opportunities on both side to be taken and the added volume and volitivity is always welcome in my opinion . Let us also celebrate the end of the longest monthly losing streak in BTC history as February 2018 comes to and end.
Now as we get I to it just a reminder that nothing I say or provide on or through this site is financial advice.
You may have noticed that several of the repositories have had very few changes in the past week or two. This is because my main focus has been on tax compliance for last year's trades and creating a forward testing program to activly monitor for the various signals I have back tested. It is my hope that in the next month or so I will see very similar results to those I got on the last three years of data.
Now sadly I will not be making this project open source as I fear in it's current state it would be insufficient for distribution. However I may turn it into a simple websocket API once I begin putting my own capital behind the trades and successfully extract capital from the markets. I will say this I built it using the gdax-pulse, gdax-assistant, and gdax modules.
Now through this development I did find a few things I needed to update..
The untarnished "golden boy" of my repositories was alas not bug free. There was an issue where the accuracy of the currentData object time was only accurate to the nearest 5 minutes. This was do to a copy paste error when getting code from stack overflow. The issue was discovered when monitoring for changes every minute and not getting them nearly as frequently as desired.
So I got a little lazy with this one and still need to finish out the unit test for some of the charts functionality. This will hopefully be coming in the next month or so as I clean up all the repositories and make sure that test data is up to date.
So there is a lot going on here. I implemented then use of funds for a market order parameter and I believe I captured a fairly accurate description of returned messages. I need play around in the gdax sandbox more to verify that the gdax-sim behaves the same way. I think I might be double taxing market orders as well, meaning I'm taking 2x the fee out at some point. I need to write some more in-depth test to either catch it or ease my mind. Lastly I have begun to implement the getOrder endpoint. This mean orders will be persistent and will be able to be queried to determine if they were filled and what their average fill price was and if they are still open. The plan is to use the order I entered with to determine the size/Funds the bot exits with. This will reduce the data needed to be stored locally to just the entry and exit order ids. And of course to test this it need to be implemented correctly in gdax-sim.
So as one might be able to tell there is a lot of work to be done. On top of this I have plans to start using evolutionary learning to extract various strategies from the market. This is risky as there is a much larger chance of over fitting, more on this to come.