#INVESTOR BEHAVIOR
Should you really check your portfolio in real time? An academic calculation says: roughly once a year.
Published on August 18, 2026
Benartzi and Thaler simulated, using monthly U.S. returns from 1926 to 1990 (CRSP database), the “prospective utility” of a portfolio under Kahneman and Tversky's prospect theory (a loss weighs 2.25 times more than an equivalent gain). They looked for the evaluation period that makes stocks and bonds equally attractive, given the observed equity risk premium (6.5%/year). Result: between 9 and 13 months depending on the method, with a central estimate of 11-12 months.
Why does checking frequency change the decision?
Stocks lose value in roughly one month out of three. Checking your portfolio often multiplies the chances of seeing a one-off loss — and loss aversion then pushes investors toward safer, but less profitable, long-term assets.
In other words: it isn't how long you hold stocks that's the problem, it's how often you look at the balance: the more you look, the more you suffer from losses that only exist on a monthly scale, not a decade-long one.
Three caveats, because they matter
- The result depends on the parameters used: with other assumptions, the authors get 8-10 months.
- This is a checking frequency, not a holding horizon: you can hold stocks for thirty years without looking at them more than once a year.
- The study relies on U.S. returns from 1926-1990; there's no guarantee the same calculation would land exactly on a year in other markets or other periods.
Why it still matters: checking your portfolio every day, or even every month, exposes you to a stream of one-off losses that don't exist on a decade-long scale — and that often push toward counterproductive decisions.
For you
- How often do you check your portfolio?
- Does that frequency help you decide, or push you to react?
Source: Shlomo Benartzi and Richard H. Thaler, “Myopic Loss Aversion and the Equity Premium Puzzle”, The Quarterly Journal of Economics, vol. 110, no. 1, 1995.
View the source →