Hacker News new | ask | show | jobs
by parhurs 3411 days ago
I'm sure the data is inconclusive for now.
1 comments

I'm also sure that a fair number of data samples could be obtained in a very short time.
Not sure, every Trump tweet that makes a positive or negative statement about a company adds one data point. So the number of samples is going to be rather small. However, whether the bot can be shown to perform significantly better than chance also depends on the size of the effect of a Trump tweet. If they produce huge bumps, a small number of tweets might be enough to show that this bot is working. Would be nice to be able to put a number on this.