Matt_Simpson comments on Raising the forecasting waterline (part 1) - LessWrong

32 Post author: Morendil 09 October 2012 03:49PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Article Navigation

Comments (108)

Comments (108)

You are viewing a single comment's thread.

Comment author: Matt_Simpson 10 October 2012 05:33:32PM *  2 points [-]

Hewitt pointed out that in general, you could do better than most other forecasters by favoring the status quo outcome.

I vaguely recall some academic work showing this to be true, or more generally if you're predicting the variable X_t over time, the previous period's value tends to be a better predictor than more complicated models. Can anyone confirm/deny my memory? And maybe provide a citation?

Comment author: gwern 10 October 2012 07:00:45PM 4 points [-]

This is a theme of multiple papers in the 2001 anthology Principles of Forecasting (a PDF of which is findable online), to give a specific citation.

Comment author: Matt_Simpson 10 October 2012 09:23:07PM 0 points [-]

Thanks! That's exactly the sort of thing I was looking for, and maybe remembering.

Comment author: Vaniver 10 October 2012 05:47:21PM 2 points [-]

I vaguely recall some academic work showing this to be true, or more generally if you're predicting the variable X_t over time, the previous period's value tends to be a better predictor than more complicated models.

These get called AR(1) models, for autoregressive 1.

Most complicated models that I'm familiar with include both the previous value and other factors (since there is generally more going on than a random walk).

Powered by Reddit Powered by Reddit