When Statistics Drop The Ball by Rob Matthews, Covenant Capital Management

There is a plethora of research available on the historical performance of trend following and it seems that extending this ‘history’ further and further into the past is currently in vogue. AQR first reported that trend following showed strong performance over the past century [1]. Not to be outdone, Lempérière and coworkers showed similar results but studied the past two centuries [2].  Most recently, Greyserman and Kaminski showed that trend following exhibits strong performance dating all the way back to the 1200’s [3].

Does this extensive and statistically significant evidence imply that trend following will continue to enjoy strong performance for the next, say, 20 years?

No. But it also doesn’t mean it won’t. In reality, just as the NFA would have us say, PAST RESULTS ARE NOT NECESSARILY INDICATIVE OF FUTURE RESULTS.

And it’s great that it’s required because it’s absolutely true. I only wish the NFA could somehow hold everyone in the social sciences to the same standard.

Statistics in markets and investing are not like statistics in physical science. Statistics in (most) physical science applications can actually be avoided if given enough information and the proper governing laws. Think of a peg board like the one shown below where a ball travels through a series of pegs and lands in a bin at the bottom. The result is a distribution of where the ball tends to land most often. This would seem like statistics at work in the physical world.

Really, if we knew exactly where the ball was dropped, and with what velocity and acceleration, the exact position and size of pegs, and the local force of gravity, and the coefficient of restitution between the ball and the peg, and the aerodynamics of the ball, etc, etc, etc – we could use the proper governing physical laws to determine the EXACT path the ball will take and where it will land (you may have seen simulations  like this before).

Social sciences have no true governing laws. This is akin to the pegs on the peg board constantly being moved without anyone knowing exactly how or why.

The truth of the matter is that no amount of statistics on the historical performance of any investment (or anything else observed in any social science for that matter) can guarantee anything about the future.

Knowing this, if you require a guarantee on your investment outcomes then you should never actually invest in anything.

Trends MIGHT cease to exist for the next three centuries, or they MIGHT perform stronger than they have in the past three. The US government MIGHT default before you cash in your bonds, or they MIGHT decide to ‘be generous’ and pay you more than you anticipated just for being a good citizen and letting them borrow your money (maybe this one will never really happen).

Any investment is basically a bet on what one thinks is likely to occur in the future, and generating statistics on the past is just one way we make ourselves feel good (or bad) about these likelihoods. Whether the markets deviate from their historical tendencies or the zebra changes its stripes is up to the future to determine.

It seems to me that if you really want to insulate yourself from this uncertainty then you shouldn’t play the game – or you should find a way to play for as short a time as possible.

-Rob Matthews

PAST PERFORMANCE IS NOT INDICATIVE OF FUTURE RESULTS

About Covenant Capital Management

“Covenant Capital Management is a boutique CTA that has been managing client assets for over 15 years.  CCM has offices in Nashville and Chicago, and employs a systematic trading methodology across global futures markets. The goal of Covenant Capital Management is to provide clients with the highest risk-adjusted returns available in the market. You can reach CCM at their website www.covenantcap.com or on twitter @covenantcap.”

When Statistics Drop The Ball