Sunday, April 27, 2014

Fallible Edges and Messy Markets

Citing Tyler Cowen's recent work, Abnormal Returns emphasized the importance of reveling in the messiness of financial markets.  Embracing uncertainty is a powerful antidote to the overconfidence biases that can plague trading.

For me, embracing uncertainty begins with market preparation.  Here's an example:

My research generally points to a higher U.S. stock market in the next few days, based on patterns that have been present over the last couple of years.  Those patterns involve sentiment, buying/selling pressure, and other measures of strength/weakness.  My chief model correlates with future market movement at a statistically significant level--and it leaves the vast majority of variance in future price action unaccounted for.

In other words, it's a fallible edge.

Fallible edge is central to how I think about markets.  Edge means that there is a reliable, repeatable, significant pattern within markets; fallible means that idiosyncratic, unique drivers of markets can always emerge and overwhelm that edge.  My model could shout "buy" from the rooftops, but that will not be helpful if military conflict erupts overseas or if a major bank announces a crisis situation.

Because edges are fallible, the outputs of any research process become hypotheses, not conclusions.  Hypotheses are meant to be tested; they are falsifiable.  Financial markets provide the test.  Based on how markets trade, I either gain or lose confidence in the forecast of the model.  If markets open and do not display net buying--or at least a drying up of selling--I then need to scan for the situational factors that might be accounting for the unexpected weakness.

A disconfirmed hypothesis is also information.

A while back, I took a car service from Connecticut, where I had been working, to LaGuardia airport for a flight back home.  The driver went past the exit that I was accustomed to taking and ended up driving around a few blocks in Queens, periodically consulting his GPS.  I finally concluded that his GPS was faulty and gave directions to the terminal.  It took quite a few minutes, however, for the driver to stop consulting the GPS and simply take the directions I offered.  I almost missed the flight.

One thing you won't find in my car is a GPS.  I study maps, internalize the main north/south, east/west roads, and then take off.  If I'm in a new place, I'll get lost.  So I'll look at a map again, figure out how I got lost, restudy the area, and then take off.

That's how I learned my way around NYC many years ago:  by getting lost often.  That's the messiness that Cowen and Abnormal Returns are talking about.  I could be the driver who relies solely on the GPS, but then I've learned nothing and I'm helpless if the GPS malfunctions.

Too many traders want a GPS for financial markets:  they look for signals and setups to tell them what to do.  But a GPS is of limited benefits if the roads keep changing.

Even the best research informs us of tendencies.  Taking tendencies for truths turns edges into liabilities.  Entertaining many hypotheses and becoming locked in few conclusions keeps traders flexible and adaptable:  ready to withdraw from markets that do not confirm expectations, and ready to pounce on those that do.

Further Reading:  Trade Like a Scientist