Sunday, April 19, 2015

Bayesian and Static Reasoning in Markets: Trading With an Open Mind

In a recent post, I highlighted the range trade in the ES futures over the past several months.  My point was that the market has been showing diminishing breadth at successive highs and also less weakness at successive lows during that range.  Generally, lengthier ranges lead to lengthier directional moves, as they are part of longer-term market cycles.  So the breakout from the current range should ultimately be a significant one.  Will the market break out of its range imminently, or will the range continue for another month or more?  Will the ultimate breakout be to the upside or downside?  Will we see a fakeout, false breakout prior to an eventual move to new highs or lows? 

My worst trading--and the worst trading I've observed of many traders--has been the result of what could be called static reasoning.  Static reasoning takes a variety of evidence, assembles the evidence into a conclusion, and then places trades based on that conclusion.  Risk taking is often a function of one's degree of belief in that conclusion.

Static reasoning is problematic for two reasons:  1) it is subject to overconfidence bias, as we take a firm stance on a view that we own; and 2) it is subject to confirmation bias, as we tend to process new, incoming information in the light of our convictions.  When I've seen traders take larger than desired losses, it's generally not been because they've held onto marginal views.  Rather, they have sized up their preferred views, stuck with those views in the face of contrary market information, and ultimately lost the position when drawdowns became uncomfortable.

I have found my best trading to result from what could be called Bayesian reasoning:  a thought process that reflects a Bayesian, probabilistic way of thinking.  Bayesian reasoning begins with a hypothesis, but it is a flexible hypothesis that updates with new, incoming information.  One's confidence in the hypothesis waxes and wanes with new information, and one's hypothesis can quickly change with new information.

With static reasoning, a market view is something you have and trade with.  With Bayesian reasoning, a market view is fluid and continually evolving.  

Trading leading up to and including this past Friday was a good case in point.  We traded firm for most of the week, with relative strength in small cap shares.  Volume had been coming down in recent sessions and, by April 15th, we saw new highs in the broad NYSE Composite Index not accompanied by an expansion in the number of stocks registering fresh highs.  With each observation of low volume and diminished new highs, my confidence in an upside breakout diminished.

Friday saw new information come into the market regarding China and Greece.  There was a strong selloff in pre-market hours.  Volume expanded, as did volatility.  New market participants were joining the fray, and they were joining with a downside bias.  That led me to sell an early, pre-opening bounce in the ES futures.  At that point, the evidence tilted toward continuation of the range and a short-term handoff from bullish to bearish control of the market.  I reasoned at the time that investors would not want to risk bad headlines over the weekend and so would be likely sellers in early New York trade.

That indeed materialized, but then something interesting happened in mid-to-late afternoon.  We had seen steady selling in stocks, as measured by the NYSE TICK.  When I ran a study of lopsided selling days such as the one in progress, I noticed a tendency for the market to bounce higher the next day or two.  At the same time, I noticed continued selling pressure in stocks (negative TICK values), but now the ES futures were holding above their lows for the day.  Selling was no longer able to get price higher.  I still liked the range-based view but the trade no longer looked great from a risk/reward perspective and I took profits.

Am I a bull?  Am I a bear?  Not really either, and the question presumes a degree of static reasoning.  What made Friday a good day in the market was the fluid transitioning from waning bullishness to waxing bearishness to waning bearishness.  We could indeed gap lower in the near term and take out Friday's low, but that's not where the odds were at the time.  Let's let the bulls take their turn and see what they can bring to the market.  Should we get a feeble rally and a lower high, there will be plenty of opportunity to resume a downside trade targeting the lower end of the recent range.  Should we get a more substantial rally, then we can update evidence for continued topping in the range or even upside breakout.

Bayesian reasoning means trading with an open mind and staying flexible in the face of new information.  Think of it this way:  we're in an ongoing conversation with markets.  In any conversation, if you stay locked in what you want to say, you become less sensitive to the other person.  A good conversationalist is a good listener, picking up on subtle cues and adjusting one's own tone and response accordingly.  In markets as in conversations, closed minds and strong views lead to tone-deaf interactions.

Further Reading:  The Importance of Emotional Creativity