ÒIncentives workÓ is a truism
that most of us discover early on in our commercial careers, but it is no more
obvious in any sector of our economy than within our trading markets. The amount of intellectual energy and
resources devoted to finding the next ÒsecretÓ correlation in stocks,
commodities, or currencies is ongoing evidence and confirmation of this
statement. Complex mathematical
models continually search for imperfect pricing in our markets because the
arbitrage opportunity, though fleeting, can generate huge rewards in the bat of
an eyelash.
Pair trading,
statistical arbitrage, or simply ÒStatArbÓ are the terms attached to this
trading practice that had its genesis back in the eighties. Gerry Bamberger is given credit for the
idea, but historical texts attribute the real success to the traders in Morgan StanleyÕs
back office trading room for turning the strategy into a giant pot of
gold. When the ÒsecretÓ got out,
banks and hedge funds were quick to jump on board, expending inordinate
resources developing sophisticated computer models to search and destroy every
profit-laden target they could find in the market.
The strategy is nearly market
neutral in its design. For
example, you must find two securities that closely correlate in price. The typical pair appearing in the literature
on the topic is Ford and GM, both in the same industry at the same stage of
development. Statistical models
calculate weighted price spreads between share values and determine where there
is a high correlation, above 80%, over time. When the spread widens, you short the high performer and go
long on the under performer. When
the market moves the stock values back to the ÒmeanÓ spread, you win both ways.
The strategy is not without
risk. If the market crashes and
both stocks go down, the short and long tend to cancel each other out. However, if the spread widens for whatever
reason and then trends in a new direction, the trade can lose in both
directions. To protect against
this scenario, risk management rules are deployed to exit trades as soon as the
set up breaks down. Quantitative
mathematical analysis, typically using
autoregressive moving average models, will forecast expectations along with
probable entry and exit points.
As the ÒherdÓ began to follow
Morgan Stanley, opportunities became scarcer, and spreads closed more quickly,
a natural result of speculating activities in a market. More focus brings more liquidity and
more efficient pricing in the form of tighter spreads, the so-called benefits
for tolerating speculators in a market.
The search then shifted to other markets. Academic studies researched currencies in the nineties, and
as retail forex trading exploded onto the scene, currencies became an obvious
target for ÒStatArbÓ speculation.
Historical pricing data can be
garnered from a forex broker, and
for programmers with time on their hands, the task was then one of finding the
proper correlations. Computer
programs, known as ÒExpert AdvisorsÓ, assisted in the search with automated
script that trading platform software (Metatrader4) used to manage positions
and orders automatically.
Since currencies come in pairs,
the search is for a favorable ÒtripleÓ, i.e., ÒEURUSDÓ, ÒGBPUSDÓ, and ÒEURGBPÓ. When correlations are over 80%, the
alert is given, and long and short tactics ensue. Activity was high some years back, but only sophisticated
traders on obscure blogs tend to share their complex approaches and results in
this remote trading arena. Results
appear to be mixed, but hope springs eternal.
Statistical arbitrage techniques
have enriched many traders and hedge funds in the past three decades. Opportunities may be scarce, but the
incentives are enduring.