Individual investors are about 50% more likely to buy a stock that they have held before, as compared to a new stock.

The above graph plots tow different probability measures for 225 Finnish stocks. The stock are sorted by daily trading volume at the Helsinki Stock Exchange between 1995 and 2009. One probability, plotted in orange, is the benchmark probability. This tries to capture that a stock bought is familiar to the investor who bought it, just mechanically. The second probability measure, plotted in blue, is the actually observed probability estimate that the stock bought is observably familiar to the investor.

A stock bought is defined as familiar when I observer that the investor has traded the stock at least once in the past one year. There are two other measures of familiarity used. First, if the stock bought is among the stock the investor encountered in his last ten transactions. And second, if the stock bought is among the last ten unique stocks that the investor encountered. The result is robust to all three definitions of familiarity.

The numbers on the horizontal axis represent one stock each. So stock#1 is the most traded and stock#225 is least traded. The orange plot is thus, as expected, decreasing. The most trader stock is more likely to be held by an investor recently and also has higher probability of being currently bought. For example, Apple (AAPL) is among the most traded stock, so any investor is likely to have held AAPL at least once in the past one year. Also while picking a stock at random, it is statistically more likely that the investor again picks AAPL. This is the benchmark probability and is thus high for AAPL. A similar probability computed for an obscure stock, say XXYZ, will be lower.

The observed probability (in blue) being much higher than the benchmark probability (in orange) implies that individuals have a much stronger affinity for familiar stocks than what would occur mechanically. And this affinity is not just for the more popular highly traded stocks but for obscure less traded XXYYZ stock as well.

Thus if an investor is observed buying AAPL, we can say there is ~50% probability that individual had bought AAPL at least once in the past one year. More surprisingly if an investor is observed buying XXYZ, we can also say that there is ~50% probability that the investor had bought XXYZ at least once in the past one year.

More details forthcoming.