Disposition to Keep Losses and Cut Winners

I have recently read a scientific research paper by Terrance Odean that was published in The Journal of Finance back in 1998. It is called Are Investors Reluctant to Realize Their Losses? The article’s main topic is the disposition of traders to close their winning positions early and keep their losing positions longer. In that particular case, they are stock traders dealing through a discount broker during 1987–1993.
To save you a few hours of reading through the terms like “heteroskedasticity“, I will just point out the main results and findings of the research conducted by Odean.
There are two points in his hypothesis:

  • Traders do close their profitable positions more frequently than their losing positions. The frequency is measured in proportion to paper gains/losses to eliminate the influence of cases where trader has overall more winners.
  • Traders do close more losing positions during December to benefit from tax savings by booking more losses. This hypothesis is of no big interest to the majority of Forex traders, so I will not discuss it further.
  • To prove his point, Terrance calculates the amount of realized gains, realized losses, paper gains and paper losses for each day when any of the 10,000 study accounts has a closing trade. Then, Proportion of Gains Realized (PGR) and Proportion of Losses Realized (PLR) is calculated using the following formulas:

  • PGR = Realized Gains / (Realized Gains + Paper Gains)
  • PLR = Realized Losses / (Realized Losses + Paper Losses)
  • The resulting PGR and PLR are numbers between 0 and 1. The closer the number to 1 the higher proportion of position closing occurs. If PGR is higher than PLR, then traders close winners more frequently, leaving more losses on paper.
    For the whole data set Odean finds that PGR = 0.148, PLR = 0.098 with a difference in proportion of -0.050, which means that traders prefer to cut their winners short and let their losses run. The author’s research also proves that such skew in behavior is not a result of:

  • Frequent or infrequent trading — PLR is lower than PGR for both groups of traders — most active and the rest.
  • Expectations of new law regulations for tax rates — PLR-to-PGR ratio is preserved if comparing 1987–1990 (when the effect of such expectations would be more visible) to 1991–1993.
  • Rebalancing portfolios — PLR is still lower than PGR if only entire position sales are counted and if no new stock purchases are made within 3 weeks of sale.
  • Rational expectations for current losers to turn into higher gains than the current winners — excess returns of closed winning positions are higher than those of the unrealized paper losses.
  • Desire to pay less commission as the cheaper stocks (losing ones are considered to be cheaper than wining ones) involve higher commission — there was no dependency of PLR/PGR ratio on stock price or stock’s current return discovered in this study.
  • Although, one may say that stock market of late 80’s and early 90’s has little to do with the modern online currency trading, it is important to note that the main finding of Odean’s paper lies not in a field of finance, but rather in psychology of trading. It seems to me that the majority of FX traders still hang on to their losing positions and try to close the profitable positions as soon as possible. That is why many prefer to trade without a stop-loss and deem it a reasonable part of their system.
    A great thing about this article is that anyone can use the presented methodology to count one’s own PLR and PGR. If you have your trading journal or a trading platform report, you can calculate your proportions of realized gains an losses. It will take about 1–2 hours, depending on how frequently you traded and how will you automate the calculation process.
    I went through the process using just pen and paper and it took me 2 hours. I analyzed 88 trades from May 12, 2008, through September 26, 2012. The resulting Proportion of Gains Realized (PGR) was 0.45 and Proportion of Losses Realized (PLR) was 0.46. The values are barely different because I did not trade very often and there rarely were more than one position open simultaneously. Still, a slightly higher PLR points out that, unlike the traders from the Odean’s study, I have been closing my losses a bit faster than I have been cutting my winners. Nevertheless, it did not help much since my total for the analyzed period is still negative.

    If you have any questions or comments about this article by Terrance Odean or if you want to share your own PLR/PGR calculations, please feel free to use the commentary form below.

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    thirty three − 27 =