I'm a big believer in sentiment analysis, but I'm really skeptical of these automated tools. The eToro project is basically just an analysis of 80 keywords, which alone sounds way too primitive to accurately capture market sentiment.
Thanks for all that info! I actually also, if forced to choose, would probably choose sentiment over anything else, although I rather feel it is a lot more intangible than any system can currently attempt to quantify (for example, all those fear and greed indices). I think almost every boom we've seen at least in Bitcoin is explained less by TA than by sentiment.
BUT what you shared only proves what I suspected. It's hard to derive sentiment from spammers, gamblers and people with vested interest (shillers, devs, paid influencers, etc.). Filter all the motivated noise and there's not much to work on. And perhaps it shouldn't filter those out since they have equal impact on retails users no?
It's hard to know. Social media shills and manipulators seem totally laughable to me, but to a newbie? They probably have a non-zero impact.
At the same time, quantifying that sort of thing from a sentiment point of view is difficult. After all, an altcoin shill (or really anyone hyping up their bags) generally represents supply: they are trying to unload their bags on other investors. They act bullish but really they want to sell. Is that bullish, bearish, or neither?
I think it's all just way too complex to analyze with such simple logic. The problem with algorithmic approaches is they aren't good at distinguishing among truths, half-truths, outright deceptions, being clouded by biases, etc.
In my experience, there is no replacement for looking at how traders are positioning themselves. Commitment of trader data (shorts vs. longs) is one thing to consider. Another thing to consider is what analysts are predicting. When everyone is piled on one side of the market, it's usually informative. When sentiment and TA align on higher time frames, we get very powerful moves.
