It’s one month since I wrote the first part of this post, I asked you about your opinion if I should make this project open source on Github. I got a good response and decided to make it available to the public use. If you didn’t go through that post can check it here.
So let’s continue.
The first concept to predict the coin choice was to use Twitter sentiments. I started with that and to train the models I scrapped the Twitter Tweets for some specific keywords. I collected the data of Two Hundred, Fifty-Eight Thousand Tweets and to check the influence level of those tweets we run a user authority test via the help of third-party API.
We got charmed and reached closed to decide which coin to buy at which time and when to sell it. We purchased the coin for current exchange price.
But all this was costing us a high computational power and server resources. It was not profitable for long-term to monitor all the keywords. At a point, I decided to shut down the project and stop it, but I already promised my friends to make it done and present to them as a contribution.
I started to google about other crypto trading bots and how they are generating signals to buy the coins. There were many sites which were providing free and paid signals I chose one of them to re-create my script.
Here this is the entire source code of the bot which I’ll explain in my next post. Please let me know if you want help to deploy that bot online.