We're currently in the BETA stage of a chat bot that provides insights to over 5000 different coins. Its aim is to alert users when certain coins are on the rise.
The system doesn’t utilize historical data or conventional data sources. Instead we use real time information from Reddit, BTC Forum and twitter. The bot connects to the APIs of the mentioned platforms. It looks for keywords in the posts of over 5000 coins, analyses the quality and quantities/ frequencies of similar posts and then reverts an alert of that specific coin with potential performance.
Be careful with all that, it will trigger the data served synchronously. You are using data based on updates from social media which are easily manipulated by some irresponsible tweets. how do you apply benchmarks in 24-hour statistical calculations? Whereas most users use historical data references? Accuracy is very important if you are providing predictive bot.