|
It seems painfully appropriate
that the guy who invented card counting in the early '60s, which he
used to supplement his income as an mathematical academic, also
started the first "quant" hedge fund in 1969. Ed Thorpe was careful
not to push his ability to beat the dealer at blackjack and other
games of 'chance' to the point that anybody wanted to break his legs
(though he was routinely "disinvited" from casinos), and he
consistently showed a profit with his hedge fund. Ed Thorpe is a
very rare thing: a man with superior quantitative reasoning skills
and good judgment.
In The
Quants,
Wall Street Journal writer Scott Patterson profiles Thorpe
and other alleged masters of quantitative modeling-based
trading strategies. This mathematical orientation consumes
Wall Street to this day, even after the big losses of
2007-2008. It is attractive, because the whole point is that math
whizzes can come up with some incomprehensible doohickey that "can't
lose," that will certainly make money. Until it doesn't.
But Ed Thorpe knows why this happens. On the Feb 1 NPR interview
program Fresh Air, Thorpe spelled out succinctly why successful
mathematical models can only be successful for limited
periods of time:
-
Modeling cannot incorporate
unpredictable events that impact the model. Or, rather,
mathematical models can predict the probabilities of
unpredictable events but cannot predict their timing or severity
with precision.
-
In an environment where
billions of dollars are chasing the latest successful investment
model, the model cannot account for huge cash flows that get
invested in the model itself. In other words, success breeds
failure by distorting the economic assumptions that the model
uses as the bases for it's predictions.
-
Success also invariably
encourages people to misapply the model to conditions and
investment instruments that are not appropriate to the model.
Thorpe's observations remind us of the
Uncertainty Principle
in particle physics,
also called the Hiesenberg Principle, that states the momentum and
position of a particle cannot be predicted because the act of
observing the particle effects momentum and position. There seems
to be a Financial Uncertainty Principle forming here: any successful
investment model will eventually implode due to its own success.
Since we can't outlaw
mathematical modeling in investing, how can individual
investors protect themselves from the tyranny of the quant? Well, a
bit of historical perspective is useful. According to the
Buttonwood column in the 2/27/10 Economist, there appears to be a
distinct cycle to equity prices. British equity prices peaked in
1906, 1936, 1968 and 1999. American equities peaked in 1928, 1968
and 1999. Apparently, investors need a generation to forget the
previous crash and pour money heedlessly into the stock market.
Which means that, sometime around 2028, you should pull all your
money out of the market and ride the trough. Remember, the one
thing that consistently works isn't mathematical modeling, it is
this: "Buy low, sell high."
This Dispatch was written with the
assistance of Brian Prioleau
|
