How often have you demanded, “Just give me a number.” You had a spreadsheet and needed to plug in one number. So you ask someone who should have it, but instead of a number you start getting a whole song-and-dance: If this happens, then it’s this number; if that happens, it might be a different number, etc. etc. All you wanted was a lousy number!

This is a well-documented tendency (see Why Can’t You Just Give Me the Number?: An Executive’s Guide to Using Probabilistic Thinking to Manage Risk and to Make Better Decisions by Patrick Leach, Probabilistic Publishing, 2006). The number you get will most likely be an average concocted from a range of values.

“Plans based on average assumptions are wrong on average,” declares Sam Savage, in The Flaw of Averages (John Wiley & Sons. Inc., 2009). Think of a drunk walking down a busy road wandering from side to side. Mathematically, the drunk’s average position is smack in the middle of the road where he might be safe, but, as Savage explains, “the average state of the drunk is dead,” not unlike spreadsheets based on flawed data.

The problem results from managers wanting a number to represent things that more accurately should be represented by a range of values. Next quarter’s sales might be W or X or Y or Z, depending on a variety of things happening. But representing next quarter’s sales as three or four or ten different values is messy.

Savage’s solution is to represent each flawed average as a set of values or a distribution string, which he calls a DIST. Savage defines DIST as a new data type encompassing a set of values that can be worked with as you would work with simple numbers.

You use these sets of values to drive Monte Carlo simulations. Instead of making your spreadsheet based on a bunch of values that more than likely are flawed averages, your spreadsheet consists of multiple Monte Carlo simulations.

When I was in graduate school, Monte Carlo simulations were something reserved for the quants, the geekiest of the geeks. Besides, such simulations consumed enormous amounts of CPU resource, which was a big no-no.

Savage insists that the use of DIST in conjunction with new technology tools changes everything, enabling managers to run complex simulations nearly painlessly in Microsoft Excel and to do so without attracting the wrath of IT. You can check out the DIST in a demo here; click on the Markowitz portfolio simulation. Savage sells a low-cost version of his Excel software here. Frontline Systems sells a bigger, pricier suite of tools based on the concept here.

You may never become a DIST fan. If you read The Flaw of Averages, however, you will think twice before you badger someone to just give you a number. ###