Among the suitable-for-blog-publication-without-risking-my-job definitions of masochism is this:

A willingness or tendency to subject oneself to unpleasant or trying experiences

So to be a forecaster, must you also be a masochist?

Few people enjoy the difficulties and degradation that go with being a forecaster, so few are willing to do it by choice. Forecasters are sometimes drawn from the top-tier statisticians and analysts in an organization.  But forecasting jobs are also used as holding pens for the lowest ranking and least experienced employees -- as a kind of corporate initiation rite with all the unpleasantries except physical hazing.

While masters- or PhD-level statistical modeling skills are not required to be a competent forecaster, it is important to have a good understanding of the business, and appreciation of statistical concepts like randomness and variation.  Mainly, a forecaster needs to know when to act, when to stay hands off, and when to stand firm against the pressures of upper management.

As we have seen in the Fildes and Goodwin research, the busywork of constantly tweaking statistically generated forecasts is largely unfruitful.  In particular, small adjustments to the forecast are just a waste of time, so there is no need to react to each new piece of information that comes in.  FVA analysis can identify where efforts are being effective, as well as where efforts are just making things worse.

Do the forecasters have authority to set forecasts, or is all their work subject to final review by a management group that can end up changing the forecast to whatever they please? Anyone charged with overseeing the forecasting staff should ask:

• If the forecasters cannot be trusted to set the forecasts, then why are they in that role?

• If the forecasters can be trusted to set the forecasts, then why aren’t they given the authority to do so?

We'll look next at the political context that envelops the typical forecasting process.