How do you interpret this data? I model trading using the mechanical price following system described below. Each simulation shows a profit using historic daily price data. When I slightly disorder the price values then in every case shorter term (20 day) tests report losses. In contrast, all but one of the longer term (100 day) trading models remain profitable when I slightly disorder the price values. SPY stock Time Constant 20 50 100 Growth Rate 0.72 0.79 1.72 After Data Randomization Factor 50 Growth Rate -1.14 0.06 0.68 MO stock Time Constant 20 50 100 Growth Rate 4.14 1.66 1.61 After Data Randomization Factor 10 Growth Rate -1.47 -0.11 0.35 GT stock Time Constant 20 50 100 Growth Rate 1.85 1.04 0.40 After Data Randomization Factor 10 Growth Rate -0.39 0.38 0.27 HRB stock Time Constant 20 50 100 Growth Rate 0.93 0.40 0.41 After Data Randomization Factor 10 Growth Rate -1.42 -0.25 0.07 DD stock Time Constant 20 50 100 Growth Rate 1.24 0.59 0.36 After Data Randomization Factor 10 Growth Rate -1.4 -0.47 -0.05 IBM stock Time Constant 20 50 100 Growth Rate 2.74 1.19 0.48 After Data Randomization Factor 10 Growth Rate -0.5 0.26 0.22 MOT stock Time Constant 20 50 100 Growth Rate 1.54 1.36 0.73 After Data Randomization Factor 10 Growth Rate -1.29 -0.09 0.09 EMR stock Time Constant 20 50 100 Growth Rate 1.31 0.75 0.56 After Data Randomization Factor 10 Growth Rate -1.19 -0.10 0.24 MSFT stock Time Constant 20 50 100 Growth Rate 3.62 1.32 0.97 After Data Randomization Factor 10 Growth Rate -1.04 -0.12 0.33 AMR stock Time Constant 20 50 100 Growth Rate 1.28 0.92 0.22 After Data Randomization Factor 10 Growth Rate -0.31 0.37 0.14 Method This price following mechanical trading system ("price breakout system") buys at the following session opening when closing price value is greater than the greatest price value of the prior 20, 50, or 100 sessions. The system sells at the following session opening when closing price value is less than the least price value of the prior 20, 50, or 100 sessions. Position size = 1 % of account equity / price value trading range of the prior 20, 50 or 100 sessions. I disorder price values by adding a computer generated positive or negative random number value multiplied by a data randomization factor to each historic closing price value. Growth rates are an average of values from 100 runs. Initial account equity is assumed to be $ 100000 in all cases. Notes: SPY daily historical price data from 29 January 1993 to 22 August 2006 (13.54 years). MO daily historical price data from 2 January 1970 to 25 May 2007 (37.41 years). GT daily historical price data from 2 January 1970 to 9 July 2007 (37.53 years). HRB daily historical price data from 12 November 1986 to 25 September 2006 (19.87 years). DD daily historical price data from 2 January 1962 to 11 July 2007 (45.40 years). IBM daily historical price data from 2 January 1962 to 12 July 2007 (45.41 years). MOT daily historical price data from 3 January 1977 to 13 July 2007 (30.53 years). EMR daily historical price data from 4 January 1982 to 13 July 2007 (25.50 years). MSFT daily historical price data from 13 March 1986 to 13 July 2007 (21.33 years). AMR daily historical price data from 2 January 1980 to 13 July 2007 (27.51 years). Growth rate is cumulative annual growth rate = (final account equity - initial account equity) * 100 / initial account equity / years.
That strategy is sort of like the Turtle strategy... I'm thinking that had the prices occured the way they did when you disordered them that traders (and investors maybe) would have acted differently. You might have negated the effects of asynchronous events like news when you disordered the data.
You might benefit from joining a Trading Tribe and getting onto the Hot Seat, to address your dramatic need for external affirmation that You Are Right, such as the posting you just made. The Tribe will also help you with written English so that you will be able to write "I modelled" and "I disordered" to describe actions that have already occurred, rather than your clumsy locutions "I model" and "I disorder". The Tribe will assist you in discovering the source of your need for public drama.
So we, once again, [sighs] hit the ignore button on a person with very few posts but a seemingly obnoxious agenda... it won't be the first time, and sadly, it won't be the last time........
I interpret this data to mean that this longer term system is more tolerant of choppy, less trending price behavior. I interpret this data to mean that this shorter term system is less tolerant and more vulnerable to choppy and less trending price behavior. These results suggest that shorter term methods, even if simulations show profitable results, might show losses in actual trading if price behavior becomes slightly less trending. I wonder if the longer term tolerance, shorter term vulnerability to choppy price behavior is true of other trading methods. I wonder if the longer term tolerance, shorter term vulnerability to choppy price behavior is true of ALL trading methods.