Well said. However, I believe there is great value in longer time periods, as you can break down the 90’s e.g. as a great time to stress test. Not only did one have various emerging markets blowing up left and right, but we had long-term capital, and the temporary “death” of value investing. I remember some saying they questioned the long term viability of the Dodge and Cox’ of the world as they “just didn’t get it.” I think the internet bubble heads did get it, just not in the way they expected! Anyway, longer time periods are very valuable, for the reasons mentioned in the previous post, but also because, frankly, this time is not different. Structural change, yes; but the importance of valuation in decision making does not diminish. These things don’t change.
I’ve thoroughly appreciated all the comments expressed in this series but if thirty years of trading has taught me one thing its to look in the direction of topics not commonly assumed at present. Volatility is a foregone conclusion for all of us and is priced into the marketplace.
What isn’t priced in is lower volatility.
It may sound absurd at present, the contrary perspective always is. For the sake of taking the other side of the volatility argument as relates to small caps is that after 4 years of tremendous volatility, could the next trend be a relatively “boring” and profitable trendlines?
It’s almost as if you’ve been reading my letters to Santa
Please bring on the 5 year cruise higher! Not sure we are there quite yet but it would be nice…
The best inoculation against future volatility is a lot of recent volatility as far as I’m concerned. Everyone sizes down and all the gamblers blow out or are too sea-sick to continue, and we get a nice run.
The best inoculation against future complacency is a bubble followed by a crash and a lot of volatility.
Interesting discussion. I agree completely that hedge funds, and many here on P123 have left O’Shaugnessy far behind. While I question the need for further data, for some of the reasons already stated, as well as simply the diminishing relevance and therefore value of the testing, the comments above sound like an argument for the max backtest. Looking at factors and models since 2001 is a way to test results under a variety of market conditions. The point in doing so is not to data mine to find the best model, but to find factors that have been consistently good, good most of the time, and good at times and not so bad at other times.
I am no good at predicting what will work in the future so I have tried what I call dynamic rankings. Testing to see what is working lately and then applying that going forward. The results are not very different from a simple diversified approach which is a lot easier.
Hi Don. What do you mean by ‘a simple diversified approach’? Is that a single sim using a ranking system that is diversified among many factors or a number of different sims?
