Oct

14

Speclist members debate the pros and cons of stop-losses from time-to-time. Don Fishback's blog mentioned this paper, Lei & Li (2009) … which I had not previously read about the use of stop-losses for stock market investors….which may be of interest to people.

Full download is here.

The jist from the abstract:

The results indicate that traditional and trailing stop loss strategies neither reduce nor increase investors' losses relative to the buy-and-hold strategy. These findings are in sharp contrast to the common belief that using stop loss strategies can improve investment returns, and the results are robust whether future returns are independent, autocorrelated, or from momentum samples, and whether we consider transaction costs or use alternative data intervals. These results also confirm our previous finding that trailing stop loss strategies can help investors to reduce investment risk. For instance, we document a risk reduction effect ranging from 28.66% to 47.08% for median and high-volatility stocks and for median and high-past return stocks under the SP strategy, when the trailing stop price is initially set at 5 daily return standard deviations below the purchase price. Collectively these findings suggest that realizing losses sooner by certain stop loss strategies can be of value to investors. This value, however, may come largely from risk reduction rather than return improvement.

Our results show that stop loss strategies do not hurt investors on their investment performance. There is no identifiable efficiency loss on the realized returns or the investment risk under these stop loss strategies. Since these strategies may provide investors with disciplines and the potential to reduce investment risk, our findings suggest a possible explanation for the widespread use of stop loss strategies in practice.

Victor Niederhoffer comments:

One wonders whether the academic paper on stops takes proper account of discountinuities in price, i..e moves that go through stop where you wouldn't be able to get out at the stop. One would need tick data for that. Also, one would state that whatever results one achieves from using stops on paper, are not valid for many reasons including the point that you never know whether a stop price has been triggered until the end of day– (a very subtle point), and less subtle, the stop price has a gravitational impact on prices making it infinitely more likely that it will be elected to ones cost than if it hadn't been elected. Usually the election is such that the price would not be hit if the stop weren't there. 

One would point out that continuous tick data would be necessary to do this properly because often the price goes through the stop and then back to a non-stop level even with 1 minute tick data, thereby making all studies using non-continuous every single tick price totally vitiable, and falsely alluring. 

One was not critiquing Mr. Humbert's summary (one knows better than to do that even if one had that in mind which one didn't) but merely pointing out a very subtle point that often prices move through the stop in continuous time, but not in time with say 1 minute tick data, and you never know whether a stop has been hit vis a vis the close until the open of the day following when the price achieved is very different from the price at the stop level or market at close level. It's so subtle I cant even explain it properly.

Rocky Humbert replies:

And I agree with The Chair's observation. The problem is reminiscent of being short (too many) options on a stock that is pinned at it's strike price on Friday option-expiry 4pm. One knows that if one hedges, it won't be assigned. And if one doesn't hedge, it will be assigned and will gap Monday morning. Kind of like carrying an umbrella– and it never rains. 

George Zachar writes:

If I may, a Schrodinger Stop is one that you place with your broker, as the physicist places the cat in the box.

But, regardless the prices you're quoted, you don't know if the stop was executed or what the fill price was, until after the session has ended….just as one doesn't know if the cat is alive or dead until you open the box.


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '5374' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search