> I don't understand why people think they know why stocks move up or down.

Inferring overly generalized and usually incorrect causal relations from extremely limited data and treating them as conclusive is a very strong human tendency; the idea of avoiding that and taking a systematic, structured, and conditional approach to assessing causal claims is fairly recent and, even among people who generally support it, often adhered to more as an aspirational principal than a consistent practice. And it certainly doesn't sell clicks the way the old way does.