Wednesday, August 28, 2013

A few things to know about Weight

I get this all the time from my risk management students: "We do a risk register and weight all the risks".

Fair enough.

But when quizzed about this, it turns out that the students aren't talking about weights, they're talking about probabilities.

Weights and probabilities are really not the same idea. They describe different things. One is about strength or influence; the other is about uncertainty. One has a 'par' value; the other is a number between 0 and 1 (can't happen; will happen, respectively) for which all related numbers sum to 1.0

Some illustrations:
  • As a portfolio manager, I may overweight (or underweight) my portfolio-- relative to a par value -- with projects of a certain class -- like semiconductor projects or projects with pharmaceutical content
  • As a portfolio manager, I assess the probability of a certain risk -- like curtailment of resources -- as an event that would impact my projects
In the first illustration, "par" might be dollars or a count, or some other metric. Thus I can talk about the weighted value of semiconductor projects in the portfolio in terms of their dollar (or count, or some other metric) influence on the portfolio as being above or below the par value.

It should be obvious, but I'll say it anyway:
  • Weights need not sum to 1.0; and in general, they do not
In the second illustration, the usual and most common definition of probability (but not the only one) is frequency of occurrence -- in effect a rate -- based on historical evidence or observation. For instance, how many times was the budget curtailed per 10 opportunities to curtail it?

If only once, as an example, then the probability of such an event is estimated to be 10%. And, by correspondence, the non-event has a probability of 90%, thus accounting for the entire event space. (You guessed it: event space = event + non-event)

Bottom line: watch your weights!

Check out these books I've written in the library at Square Peg Consulting