Oct

9

Stops, from Steve Ellison

October 9, 2007 |

Wrestling with the question of where to place a stop on a position intended to be held for a day, one approach I took was to start with the probability of being stopped out.  If I wanted, for example, a 10% probability of being stopped out, I could look at the last year's daily lows, express each as a percentage decline from the prior close, and find the 10th percentile of these daily maximum adverse excursions.

In datamining with S&P 500 data from the 1980s and 1990s, I found that the tighter the stop, the better the total return.  Had one decided at the advent of S&P futures trading in 1982 to buy the close each day and sell the close the next day with a stop placed so as to have a 75% probability of stopping out, based on a one-year lookback — resulting in stops as little as 0.1% below the prior close — the theoretical return without considering commissions, slippage, rounding, and gaps down would have been an incredible 7,918% by the end of 2000, versus a 528% return for buy and hold.

Alas, this effect disappeared completely after 2002.  The same strategy would have returned only 14% from the end of 2002 to last Thursday, versus a 64% return for buy and hold.


Comments

Name

Email

Website

Speak your mind

5 Comments so far

  1. AlfaMike on October 9, 2007 7:43 pm

    Personally I’ve found small stops to be frustrating and ineffective. Flushes trigger them and then the stock goes back up. How about using the IBD approach, buying up trending stocks on pull backs to the 10 week moving average? And setting a stop significantly below that. Seems more realistic.

  2. lawrence schulman on October 11, 2007 1:09 pm

    The bull market began in 2002. Tighters stops should not be placed in a bull market would seem to be the conclusion drawn.

  3. Denise Shull on October 11, 2007 8:06 pm

    In this age of ever increasingly connected global markets, what is a day?

    Having asked that rhetorical question I can only add that this is the most difficult question in trading.

    Aside from the infinite variations the mathematical analyses present, there is the human reality that almost no matter which strategy you choose - you lose on one of two counts. Either you didn’t hold onto a winner or you gave back a significant amount of profits.

    Either way, this invokes the emotional side of the joint probability distribution - which at a minimum typically leads to further tinkering with the math and the game starts over….

    I would submit that you might ask yourself the question “What can I live with?” By always keeping the answer in mind, you build the psychological capital that allows you to more efficiently answer the monetary capital questions.

  4. Craig Bowles on October 12, 2007 6:46 pm

    Enjoy your postings. It looks like the daily highs and lows show a lot less volatility since it hasn’t been working. Maybe some many folks are playing the small reactions? Pretty odd. I used to have a book about that said a good daily trade is when the market gaps up, sell short if the previous day’s range is re-entered and the opposite on the gap downs. Maybe your system would still work if you just tighten the stop on the gap moves and set an entry point going the other way if the move fails. Maybe using the 3-day Arms index average with seasonality could tell weather to initiate the daily buy or sell signals.

  5. TWI on October 13, 2007 7:38 am

    I have found that on short term trades where I expect to be in and out within the session then a fixed pip stop works far better than trying to fiddle about with volatility adjusted stops (these are great for longer term positions). With S&P I also find buying strength or selling weakness ineffective. The market clearly mean reverts, this can be simply demonstrated by selling new weekly highs or buying new weekly lows and holding until the trade day + 3 open. It shows a reasonable return, better than anything buying highs or selling lows. Oddly enough the european indexes work better with breakouts.

Archives

Resources & Links

Search