Mark up a new one for the back-testing books

Back-testing proponents proudly show the empirical base of their approach. No parametric fitting or theoretical assumptions here. “We use the actual history of the markets to assess our risk and define our strategies.”

However, the debate that typically comes up is:

Back-testing over which period?

Do we include the dotcom bubble and burst? What about Black Monday? What about the oil-shocks and stagflation period that “will never happen again because we understand inflation”? What about the other Black Monday (and Black Tuesday and Thursday and sometimes Friday) of October 1929?

Nassim Taleb will happily say that we should, of course, include all those and the events that have not yet happened. He’s a little more quiet on how. (Ok, that’s unfair, he recommends in absolute risk management which strictly speaking doesn’t need to model the worst possible scenario.)

Well, the events of the last year and a bit will change all those back-testing models. Many strategies and positions that seemed relatively safe in a 2006-vintage model will look frayed and dismayed. Northern Rock belongs to Mervyn, Bear Stearns down and out, Fannie Mae and Freddie Mac nationalised, Merril Lynch sold to Bank of America, Lehman Brothers declaring Chapter 11 and AIG down around 80% over a year.

Mark my words, the end may be nigh, but there’s more to come before we get there.

It’s a great time to be living in the markets. Especially if you’re in South Africa with a hefty portion of cash in your portfolio.