As I’ve mentioned above, there are two modules of Machine Learning into that platform: price prediction and news analysis. Let’s talk about first.
Obviously, I can’t go into very detail about the process of buidling models as it was a commercial project, but, believe me, I will give you and idea to try 🙂
Despite all the articles that suggest just collecting the historical price data, I went further and collected far more than just “numbers”. The dataset, that I used to train the model are news for specific date, views, likes and many more. There are more than 50 features participated in building the model. My main goal was to collect all, literally, all info about the coin so no data will escape from me.
Based on AI Price predictions, the developers then used that info to provide users with crypto signals: entry point, SL/TP etc. All across different coins.
The second module that uses Machine Learning is real-time news analysis. I collected the data as using APIs as Scrapping. The collected raw data was then transferred into text-processing pipeline and matched with time series to give better understanding of what was social environment at that current point of time for the specific coin. This data was used as for time series forecast as for market sentiment prediction for tomorrow.
The are some more AI features in the app like chat bot that can answer any questions about the platform and the crypto trading but main features and those I talked about above.
In the next articles, I will disclose you more about how AI is integrated into real products. For now, I hope this article has been helpful to you 🙂