normative
Out-of-sample holdout testing as a remedy
To guard against spurious patterns, one should form a theory first, then test it out of sample, in different markets and environments, and tweak variables that should not matter; if the result survives changes to irrelevant variables, the finding is more likely real, but if it fails, the analyst is probably fitting noise.
normativepending
Speaker
JesseEvidence Quote
“if you can tweak variables that you know don't matter, that's always great... if it still survives, that's going to increase the probability that's true.”
Created: 11/11/2026, 6:43:50 PM
Your notes
Loading notes...