It is, or should be, an expected result. But apart from that, don't believe what is posted here, test it for yourself and see.
It would help, but it isn't sufficient. Any data the tester/designer has seen or used in any way is already "in sample". However, that is a long discussion....... Not "is" profitable. "Would have been" profitable.
This guy said he was able to start a new career with this idea. Wonder if it is still working? Regards, C 40yotrader Registered: Oct 2002 Posts: 73 10-29-02 04:12 PM Re: 40yr -------------------------------------------------------------------------------- Originally posted by trader388 can you tell us a little about your system? -------------------------------------------------------------------------------- It's a volatility breakout system. It measures the range for the past 5 days. Then it figures out the lowest of the 5 ranges and adds/subtracts a percentage of that number from the next days open. It also uses a sliding envelope for the exit and reverse point. For example, if the breakout point is 10 pts., then the initial buy point is the open + 10 pts. If the market opens and drops 5 points, then the buy point slides down and becomes the open + 5 pts. Then if the market rallies above the open and hits the buy stop at the open + 5 pts. it enters a position. The sell stop for reversal would become the entry - 10 pts.. If the market went higher by 3 pts. and then turned, the sell stop would slide up so that it's the buy entry point - 7 pts. Once today's range is within some percent of the lowest 5 day range, a trailing stop is used for the exit of the current long or short position. The trailing stop is a parabolic stop using a factor. The factor is determined by the slope of the move since the entry to find a best fit so that it will exit when the trend starts decaying. Once exited, there are no more trades for the day. One filter it uses is that yesterday's range must be less than the previous days range to initiate a trade. The money management uses a combination of a base number of contracts and adjustments. The base is figured from the average range of the past ten trades versus the previous ten trades. If the range is declining, then the number of contracts is increased. Likewise, if the range is expanding, then the number of contracts is reduced. The adjusted contracts traded is based on all kinds of conditions like 1). If in a drawdown, don't reduce the number of contracts traded until a new equity high is reached, 2). Increment the number of contracts to trade if the last trade was a loser and lost more than the average of the past ten winning trades. 3). If the average winning of the past ten winning trades drops below $2,000, then compute how many additional contracts are needed to keep above this number and add the adjustment on the next trade.... I don't thnk the system is any big deal. All the real profits come from the money management. I needed all this stuff to make sure it would reach my goal of 150k/yr. in income with short drawdown periods.
The article is intended to only be an example of how one can use a systematic approach to developing a trading system. As such the test period was only 6 months, during which the day of the week test showed Thursday's to be overwhelmingly poor. As the system was being constructed using data from this 6 months only then the Thursday rule had to be included as that was the point of doing the test. In out-of-sample testing for 18 months it can be seen that Thursday is still the poorest performing day of the week but no where near as bad as during the actual test period. But if the out-of-sample data is then taken into account and the Thursday rule scrapped then the out-of-sample data merely becomes additional test data! The point of out-of-sample testing is to check that the whole system is not over-optimised and consequently falls apart completely on data that wasn't used to develop the system. The article does make it clear that there are only 22 Thursday's during the test period and it may not be statistically reliable. Overall, though, the Thursday rule only has a marginal effect on the results of the system - It certainly doesn't turn a losing system into a winning one!
Sid, Thank you for posting your thoughts on the Thursday rule. I am trying to code the system into Wealthlab and test it for myself. Another Wealthlab enthusiast has already done it and confirmed that the system is profitable for the current March ER2 contract. Please don't let critics deter you. Most of them mean no ill, and perhaps to help. Keep up your great work. Maji
The system had 3 winning trades out of 3 last week. Could have been 4 out of 4 if it wasn't for the controversial Thursday filter! 2005 system results. If anyone is interested an excel spreadsheet containing all the results from January 2003 onwards can be downloaded from the site, here.
It's a fascinating and excellent article, and a pleasure to read; thank you. Have you ever looked at the prospects for doing more or less the same thing with the Dax index? I have been told (by several people, actually) that it's the most profitable index for this style of system, but that the best "opening range" for the break-out is much shorter than is the case for any other index. Do you have any views/ideas/opinions about this? Or even any experience of it? I'd be very interested to know.