You are right, detrending is a good start but is not sufficient. If you're interested the next step would be to look at how volatility relates to price. Ninna
Do you mean that if you normalize prices properly, oversold conditions may turn out not to be so and the same for oversold? However by normalizing prices you do not change the number of up or down days. Neither you change their relative magnitude. I may be wrong, but normalization will result in the same RSI values if the same normalization procedure is applied to both down and up days.
I did not mention normalization of prices. I use Connie Brown and her insights regarding the RSI. I have discovered and coded the Composite Index, which is a companion to RSI. I will not give it away because she has the formula in her book, and i will not dishonor her in that fashion. If you ask me to supply a chart, I will, but I will not ever give away the formula to the Composite, because it is there to be had, if you are not lazy.
I fully understand your concern. However, that is an artifact of a small sample space. With a large sample space that becomes less relevant as the strategy's results approach its true probability. For example, when the number of trades taken is sufficiently large you can randomly remove taken trades and the winning %,profit factor, etc remain relatively stable. The same observation can also be made by taking out trades in a more deterministic fashion such as remove every 3rd trade etc. Still the important metrics remain stable. Failure to remain stable would point to a systemic issue in your results and would be cause for concern.
This is what I aim to prove/disprove in this exercise. The results I am publishing is in stark contrast to your stated observations. Buying when first crossing into overbought and holding for the next several bars was a losing proposition in all tests. On the other hand buying when crossing BELOW a threshold such as into oversold worked out very well and showed a statistical edge when compared to random entries. How do you reconcile the differences between my results and your observations?
"The reason, The Composite Index looks a lot like my work is because it is my work. All Ms. Brown did was change the name of my "CFG" Oscillator. My work was copyrighted and reistered with the US Copyright Office back in 1990. And as for my Cardwell Positive and Negative Reversals Patterns in RSI I have been working with them since 1978 and have taught them in my seminars and courses since 1989. I still make my courses available (RSI Basic and The RSI EDGE) and you will find information on my website -- cardwellrsiedge. The RSI EDGE Course was first offered in March 2002 and has my CFG ( what Ms. Brown calls her Composite Index) and is included as part of the course. I also have other indicators which I have developed over the years. Any of my other work and research that I have developed since The EDGE I forward to my course students as part of their course. Each student has my cell number as well as my direct email address. I stay in touch with my students and many send me charts of current markets with examples of what they first learned in my courses. I guess that's why over 70% of my students have been referred by previous students and we have students and clients in now 28 countries around the world. I am here and available to speak to anyone who would want to get in touch. Check out the site and send me an email. i look forward to stopping in here every so often and hope to share some insights and my thoughts. For now be careful in the stock market. I think we will be testing the lows again. This rally since March is just a correction in still a long term downtrend. Thank you and all the best always to everyone. Andrew Cardwell" http://www.trade2win.com/boards/tec...tance-browns-composite-index.html#post1205014 BTW he was wrong about #2 when he posted in Aug 2010
Wow! Learn something new every day! Well, at least the OP is aware that Andrew Cardwell is the authority on the RSI. She does mention that. On the other hand however, how is allowed to continue to publish her books with stolen info, why does he not sue?
I was not replying to you Then, if you are using any technique available in any book, you are certainly not having an edge. I know people that constantly fade the signals of that formula and make money at the expense of fools who read some technical analysis. They have even devise projections to know when the fools will be placing tardes based on that and similar indicators.
Actually you change the relative magnitude because your normailzation factor isn't constant. That way you can remove the fat tails. Imagine you're at a coca-cola factory and you see all bottles coming out. You know that the content of each bottle is normally distributed with a mean 33 cl and std 1cl. In other words for 67% of the bottles the content is 32-34 cl and for 95% it is 31-35cl. Now imagine that someone lets you bet on whether the next bottle you see will have less or more content than the previous one with a r:r 1:1. If the previous bottle had 36cl or 29cl it would be an easy bet right? In trading I'm trying to set up the same kind of betting proposition. Ninna