LunarCRUSH trains machine learning algorithms to gather and access data to return back sentiment on that data. We then cross train the machines in a never ending cycle that allows for the most accurate return of data in the broadest data set available.
We hand train our machine learning based on crypto-specific data. We do not utilize standard natural language processing sentiment algorithms as we do not believe they correctly access the language used within the cryptocurrency community to express emotion.
The language used in the cryptocurrency market can sometimes differ quite greatly from traditional stock markets.
For example: "Monero is currently mooning." Is a Bullish statement. You probably will not hear an analyst on CNBC say, "Duke Energy is mooning folks!"
At LunarCrush, we utilize machine learning to hand-train content for it's Bullish or Bearishness.