(Bloomberg) — Hopefully the algos that read the headline above have a sense of humor.
Because this is a tale about the traders whose strategies are influenced by actual human emotion coded in their DNA, not coded into millions of lines of C++. And that is exactly what we non-robots do:sell the portfolios winners and losers way more than the rest of our stocks, according to research by Samuel Hartzmark,
assistant professor of finance at the University of Chicago Booth School of Business.
Hartzmark called it the rank effect and said it can be seen in trading by individual retail investors as well as mutual fund managers. The phenomenon exists even after controlling for things like company-specific reasons for why the stocks are being sold, or selling related to tax strategies. According to Hartzmarks paper: the effect exists purely comparing across investors or across investors with similar holding periods.
On days when individual investors decide to sell, they have a 31 percent chance of dumping a position with the highest return in the portfolio and a 26 percent chance of unloading the one with the lowest, compared with an 11 percent probability of selling one in the middle, according to the study. Mutual funds also show tendencies to dump the winners and losers more than the rest.
Alphabet Effect
One explanation for the phenomenon may be that brokerage accounts list holdings in order of returns and investors may fixate on the best and worst performing stocks, Hartzmark wrote. To bolster the case, he notes that a similar effect exists with stocks at the beginning and end of the alphabet, another way a list of positions can be sorted.
Hartzmark is not the only one to notice the alphabet phenomenon. Ana Avramovic of Credit Suisse Group noted in a July report: stocks that fall in the beginning of the alphabet tend to also have better liquidity characteristics such as average daily volume and bid-ask spreads.
Anyway, you can read more about Hartzmarks study at Booths Capital Ideas blog or troll it up with the other quants over at Larry Tabbs Quant Forum.
One caveat to the study is that the retail brokerage trades examined happened way back in 1991-1996. That corresponds pretty closely to the rise in popularity of professional wrestling in the U.S., so maybe we were just collectively more stupider back then.
