>we should juxtapose the stock market and see how things correlate.

You shall find this relevant to your interests: https://www.sciencedirect.com/science/article/pii/S187775031... (by a former colleague of mine) they claim to get signals that lead the DJIA by 1+ days.

It's a big mistake to assume the people of twitter are a good sample of the people of the US.

I think the bigger mistake is to think even if it was a good sample it could gauge people’s mood - incentives for posting to social media are either happy or outrage. You will never get a silent majority “just okay” or “neutral-low” by averaging between ecstatic and outraged.

"The squeaky wheel gets the grease"

I think the fact that two of the top ~10 happiest days in 2018 were attributed to a Korean pop star's birthday is excellent evidence that people of twitter are not, in fact, good samples of the people of the US.

Also, the sadness when Trump was elected despite relatively even vote totals tells you that Twitter skews to the left.

80% of the population owns 6.7% of the stock market

but really, what about this sine wave? is this some kind of an error in data?

Looks like a typical seasonality curve. Things tend to repeat at the same period of the year (say, happy in summer, sad in winter). You can see similar patterns at different time scales (weekend vs week, day vs night etc)

but there is this huge ~4 yr cycle