26 October 2014

You Don't Have to Believe Me

Recent musings, motivated by continued discontinuity between real world finance time series and micros/quants insistence on predicting past the data, have been devoted to this problem. The world is not linear (Dr. McElhone), and is mostly driven by incentive modifications rather than historical data. Those that believe otherwise still think The Great Recession was caused by the Trilateral Commission attempting to impose a New World Government. Baloney, of course.

On the other hand, some academics understand that time series prediction is flaky in actual practice.
There are at least four sources of uncertainty in forecasting using time series models:
...
4. The continuation of the historical data generating process into the future.

QED

No comments: