Thursday, June 12, 2008

Stop It!: Stop-Loss Results for Bullish Scenarios

In my blog on May 27th we looked at different types of common stop/loss strategies and discussed when and how to use them. I also mentioned that I was going to test out several of the more common strategies under various market conditions. After a software snafu on the part of the software vendor (if I'm to be a beta tester, I'd like to get something for my efforts!--grrr), I finally was able today to finish my long simulation.

The Set-Up
What I did was to look at three different portfolios of 10 stocks each--one composed of stocks priced less than $10, one composed of stocks priced between $10 and $30, and the other of stocks priced between $30 and $100. My simulation consisted of finding stocks breaking to new 250 day highs, which is a good strategy in a bull market. I ran my simulations at two different times when the market was on a bull run, from 9/26/06 to 2/20/07, and from 3/19/07 to 6/4/07. (Simulation and management account information are described at the end of the article.*)

The following stops were used: Trailing stops between 5%-50%, Gain/Loss stops with Gains set at 1000% (literally no stop on the gain side) and losses varying from 5%-25%, a ratchet stop, and finally, no stop at all. I would have loved to test out the parabolic SAR, but my program couldn't do that and it would have meant weeks of manual input. (As it was, this took me long enough.)

The following table shows the results. The first row of each simulation shows the profit or loss over the time period. The second row is the number of trades which each trade being equal to a buy and a sell for a total of two commissions (what is known as a round-turn in futures parlance). The win/loss percentage is in the third row followed by the maximum portfolio draw-down in the fourth row.

The Results
The first thing to glean from this table is that it could use more data. While we can't pinpoint the exact best stop/loss to use in each case, we can note certain tendencies.
Low-priced Stocks (<$10): In the case of small stocks, we see that trailing stops didn't do nearly as well as the G/L stop. An absolute loss of 5%-10% seems to work best while also keeping the number of trades to a minimum. Note that the use of trailing stop losses involves significantly more trades, and although commission costs are included in the profit/loss figures, executing this number of trades would take a lot of time on the part of the trader. Trailing stops also resulted in larger portfolio draw-downs. The ratchet stop did well in one scenario and poorly in the other. The reason for the poor performance was just bad luck--selling one stock at one particular time had the effect of the program selecting poor performers down the line (sort of a snow-ball effect). (This shows you why you need more than two simulations to get reliable data.)
Mid-Priced Stocks ($10 - $30): Contrary to the small stocks. the trailing stops fared better here especially in limiting draw-downs in the second market scenario.
High-Priced Stocks ($30 - $100): Here, absolute losses between 5%-15% out-performed the trailing losses, with the 10% G/L edging out the others. **

It's clear that we need more data points to identify which stops to use. Setting stop/losses is as much of a science as it is an art, and setting the most appropriate ones depends on many variables, such as market type, stock price, stock selection criteria, etc. The above table makes it clear that setting the right stop/loss results in better returns, but it's also interesting to note that not setting a stop/loss isn't that bad, but one could expect that since we're buying strong stocks in a strong market which is probably is not the case for other market scenarios. One thing that surprised me was how well the ratchet stop did (except in one instance). You can also see how poorly the 5% trailing loss did in general and the inordinate amount of trades it requires.

The bottom line here is that if you have to pick one type of stop to use, I'd go with a strict percentage loss between 5%-15% followed by a ratchet stop. Or, you can fly by the seat of your pants--use no stop/loss criteria and exit when your stock breaks some type of resistance or starts to sell off. (See “Other Stop Loss Methods” at the end of my May 27th blog.)

When I have another chunk of time, I'll look at setting stops under bear market conditions.

*All accounts began with $100,000 initial investment in a 50% margin account. Trade commissions were $9.95/trade. No dividends were collected and account interest was included. Each portfolio maintained 10 stocks that were automatically selected from a list generated according to the parameters of my break-out strategy. The strategy parameters were as follows: 1. Only stocks that traded on the major indices (NYSE, NASDAQ, and AMEX) were considered; 2. Average daily volume must be greater than 200,000 shares; 3. The volume on the break-out day had to be greater than the average daily volume; and 4. A stock break-out was defined as one breaking to a new 250 day high. Closing day prices were used in all calculations and no duplicate holdings were allowed.

**Blog Update (6/13/08): I changed the wording of the stock categories from Small Stocks to Low-Priced Stocks, Medium Stocks to Mid-Priced Stocks, and Large Stocks to High-Priced Stocks. Prof. Pat pointed out to me that my original wording could be construed as company size which is not always the case, so I changed the wording to more accurately reflect reality.

No comments: