The MacArthur Foundation gave out its latest batch of “genius grants”  yesterday, recognizing  “exceptional  creativity in their work and the prospect for still more in the  future.” One of the winners is a CalTech economist  most famous for calling traders idiots. In 2010, Dr. Colin Camerer co-authored, “Using  Neural Data to Test A Theory of Investor Behavior: An  Application to Realization Utility.”  It is his most downloaded  paper, according to the St.  Louis Fed. Camerer and his co-authors found that the “realization utility” model of  investing is way more prevalent than it should be. “Realization utility” describes the phenomenon of a given trader being more  prone to taking “realized gains,” or immediate profits, than to allow “paper  gains” to linger on their theoretical balance sheet. As they put it: “[The  ‘realization utility’ trader is] keen  to realize capital gains as soon as  possible and to postpone realizing capital losses for as long as  possible.” For a significant number of traders, a part  of the brain that should not be given the keys, so to speak, to controlling  trading decisions ends up doing so anyway more often than it  should. When that happens, this results:

“… there were a total of 495 occasions in which  our subjects realized gains, and that most of these decisions were suboptimal.  Given that stocks exhibit short-term price momentum in the experiment, it is  generally better to hold on to a stock that has been performing well. This  explains why most (77.9%) of subjects’ decisions to hold on to winning stocks  were optimal, and why most (67.5%) of subjects’ decisions to sell  winning stocks were suboptimal. Similarly, in the experiment, it is  generally better to sell a stock that has been performing poorly. This explains  why most (79.2%) of subjects’ decisions to sell losing stocks were optimal,  while most (80.3%) of their decisions to hold these stocks were suboptimal.”

Here’s the chart:

 

The finding “stands in sharp contrast to  the prediction of a simple rational trading  model in which subjects maximize the expected value of final  earnings.” Ouch. Camerer’s most cited paper, according to the St.  Louis Fed,  is his 1999 work “Experience-Weighted Attraction Learning In Normal Form  Games,” which proposes an entire new way  of looking at how humans learn, called “experience weighted  attraction.” Camerer and his co-author Teck-Hua Ho found this concept captures the  phenomena of subjects combining belief learning and experience learning.Camerer and his fellow “geniuses” get  $625,000 paid out over five years — not chump change to an  academic!