[Houshalter]

A few months ago, before the current wave of hype started, I got the historical price data on bitcoin and ran Eureqa on it, which is a program that tries to find the simplest possible equation that fits the data. Not for any practical reason, I was just testing the software. I extrapolated the result forward in time a few months ahead to see what it predicted, and it showed it increasing in price exponentially.

I was disappointed because I thought that was ridiculous and that I must have overfit somehow.

[NamTaf]

You didn't overfit. You fell in to the trap of assuming that past performance dictates future performance. It does not.

[frisco]

> I must have overfit

You did overfit.

[Houshalter]

I don't think it was overfit, though just historical data isn't anywhere near enough information to make accurate predictions about things. I just find it funny that it wasn't as far off as I thought it would be.