Saturday, December 27, 2014

Parallel Processing in Trading: Finding Meaningful Patterns

In yesterday's post, I described a process of reading in parallel:  reading many books at a time and gaining insight into a topic by juxtaposing the views of many authors.  Parallel thinking is central to pattern recognition:  in simultaneously processing multiple events, we are able to discern meaningful patterns connecting those events.

Take a look at Rob Smith from T3 describing his eight screen trading station layout.  There are many charts on each screen, grouped by sector, stock type, etc.  He then can refresh all the screens at once to view many other stock groups based upon screening criteria.  Each stock can be viewed across multiple time frames.  As Rob points out, he monitors all of these screens throughout the day to get a sense for when opportunities are "lining up".  Then Rob will go "around the horn", looking at every stock in the SPX to detect emerging trends or moves out of the ordinary.

What is noteworthy in the video is that Rob is processing much more information much more rapidly than the average trader.  He is reading the market much like I am reading books:  finding themes by processing multiple sources in parallel.  Instead of examining one stock in detail, consulting myriad indicators and chart perspectives, Rob considers many stocks and finds patterns that cut across them. 

If you jump over to SMB, you'll notice that their traders are utilizing tools that filter stocks based on liquidity and volume and then track promising candidates tick by tick to detect unusual volume or order flow patterns.  The technology acts as an extension of the traders' parallel processing, reducing an impossibly large array of intraday data across stocks to a manageable universe of "in play" opportunities. 

Serial processing is common among investors:  digging deeply into particular subject areas to arrive at unique analyses that become trading opportunities.  An example would be scouring the wording of Fed statements and speeches of Fed officials to discern shifts in policy.  Parallel processing is less about deep analysis and more about rapid synthesis.  It is more common among high speed traders:  finding patterns in market action that reveal shifts in supply and demand.

Most of us possess thinking styles that are our unique blend of parallel and serial processing.  An important source of failure for traders is attempting to adopt trading styles that do not make optimal use of our cognitive strengths.  An important source of failure for trading firms is failing to assess cognitive strengths as part of the hiring process.  The myth continues that trading success is a function of personality, while evidence strongly suggests that personality can accomplish little in markets if the right brain wiring isn't in place. 

Emotions are a problem in trading only insofar as they may nudge us from our cognitive strengths.  When trading becomes challenging, higher frequency traders should push themselves to look at more things and feed their pattern recognition; lower frequency investors should push themselves to think more deeply about what they're doing and why.  Bad things happen when active daytraders respond to challenge by slowing their thinking and when investors become more speedy.

But of course, we don't hear about any of that from would-be trading mentors and coaches.  They tell us to "trade our plan".  

Whatever.

Oil was weak most the day on Friday.  High yield bonds underperformed stocks and then saw decent selling late in the session, as sell programs took stocks off their highs.  That was a piece of pattern recognition from yesterday's trading.  When all that began to unfold, my long ES position came off the table.  Not all patterns are meaningful, but that was a movie we've seen before and I wasn't in the mood to replay.  If you're only looking at the chart of what you're trading, it's tough to see those intermarket patterns in real time.  Markets are always talking to each other; it can be very helpful to join their conversation. 

Further Reading:  Data Rich, Information Poor
.