trading eminis with cycles

Discussion in 'Technical Analysis' started by tokyotrader, Aug 27, 2005.

  1. I am new to cycles(time series analysis) and was inspired by an old poster who used them to accurately predict emini movements. Is anyone here using cycles to successfully time the ES & YM and if so please share your experiences.

    Here is a screenshot of some cycle analysis I tried(hope this attachment works). The purple line and the black lines were projections from the beginning of the week. Although of course this is now in the past, the settings I am using are in a way being backtested from the anchor point I am using to see how well the projections match what actually occurred. Looks promising with the exception of Friday morning.
  2. gnome


    You've got a strange post. I've never seen anyone overlay any kind of "time" chart over PnF charts... that's probably because time is specifically ELIMINATED from PnF consideration.

    I've been trading LONG time, and I've never found anything LESS RELIABLE than "cycles".

    It's most likely that any effort you put into incorporating cycles into your trading will be frustrating and ultimately worthless.
  3. Thanks for your post. You are right, time intervals are not constant for PnF. I have seen something similar work well, hence my attempt to explore that area.

    Might I ask what type of analysis you did with cycles? Also, you concluded that cycles were unreliable. Do you know of traders who have successfully used them? I am certain they are out there. I am interested to hear from both sides. Those who did not find them useful and those who did. We don't need to argue who is right or wrong, but let's open our minds to the possibilities.
  4. gnome


    I've looked at cycle considerations off and on over the years. Sometimes it was because somebody said he had a clue and I wanted to check it out. sometimes I just thought I should revisit the concept to see if I could capture anything.

    Bottom line...... Nada. Nothing. Zilch. Now, whenever I hear someone mention the word "cycle" I presume he is delusional and immediately hit the mute button.
  5. What is it you are using to determine next cycle ? Who was an old poster ?
  6. Allow me to save you some time. I studied cycles and how they related to over 50 Markets (Cash, Equities, Indices, Currencies & Commodities) for 2 years and can tell you there is absolutely no consistency in using them.
  7. slacker


    I agree with Charlie Dow and gnome cycles are not helpful. I spent a 'lot' of time going through all of the Ehlers, J.M.Hurst, and using digital signal routines trying to find the holy grail in cycles.

    There is nothing in phase, amp, or frequency that will help tell you what the next bar is going to do. Nothing.

    I do my work in C++ using Numerical Recipies routines for the signal processing code. In spite of spending a lot of time in this area with no success I returned to cycles recently to see if just using 'phase' alone could help me get a slightly better entry or exit. I am using Interfactive Brokers API and wanted to improve my automated entry and exit process just a tad. Once again nothing after some fairly complicated code slinging.

    For prototyping I use Tradestation. You can find several good posts in this area on the Tradestation forums. Cycles are very attractive to engineers who have any kind of digital signal processing background, or like to play numbers games.

    Best work in this area for trading apps is "Joint Time-Frequency Analysis" by Qian.

    As to your Point and Figure chart with cycles you are going in the wrong direction (IMHO). You are going to lose a lot of good information....

    You can also look up the MESA software and trading systems led by Ehlers and look at system results posted in MESABonds is have a good year followed by several years of 48% drawdowns. MESANotes also well underperformed the S&P for the past 3 years. The results don't get any better if you go back 10 years or more. You don't need 'cycle' processing to get these kind of results.

    Another thread with more info:

    Just some thoughts on a lazy saturday morning.
  8. I concur. Cycle work is attractive to a lot of engineers who get into trading, because the price time series lends itself to the same type of analysis used for signal processing. Hence the use of fourier transforms to extract dominant frequencies, etc. Unfortunately, whatever cyclic component that exists in a financial market tends to vary over time. In other words, this week's cycle won't work next week.

    If you are interested, Ehlers has done a lot of work in this area and sells software and trading systems. Check out his articles in TASC. Persoanlly, I think it is a dead end.
  9. Hi AAA,

    Very accurate observation. About Ehlers 'work', I have followed this 'market engineer's endeavors for years. Besides commercial laudations of its merits, I HAVE NEVER come across of any demonstration that this stuff has any value whatsoever. I can't understand how, to my knowledge, so many articles have been allowed to be published without even a minimal verification through backtesting. I tend to concur with your dead end opinion.
  10. Holmes



    don't let anyone discourage you examining something that you believe in. Perhaps there is an angle that no-one has looked at and that is working. :D :D :D

    I found something that others also said would never work, yet it works better than what they have. This is because I took a different (non conventional) approach to it.

    May the force be with you.



    Shamelessly reproduced here from fellow ET'r travis: (thank you)

    #10     Aug 27, 2005