Adaptive Trading Strategies

Discussion in 'Strategy Building' started by jalsck, Sep 4, 2010.

  1. jalsck

    jalsck

    I'm not 100% confident that I understand what you mean by bar-after-bar. It is similar to the overlapping block approach (bad description I know) except the optimization period is always trailing the current bar, so all signals that are generated are out-of-sample. Naturally, this assumes that the parameter ranges have not been optimized using the data put aside for the trading simulation.

    Did you have any securities in mind? I am happy to do some runs for you. Given that you have specified SMA(5) > SMA(30) I would set the parameter ranges for the first SMA from 2 to 8 and the second from 20 to 40 so that the system has the opportunity to adapt. Also, is this a long only system or do you want to go short when the shorter SMA crosses below the longer SMA?

    Best Regards,

    James
     
    #31     Nov 10, 2010
  2. it is realy not that hard to understand why this aproach is considered to be optimizing rather than adapting. on this aproach of yours have systems been built allready, a tradestation system that sold for $5k from someone that i could provide his name later to you (i forgot it) comes to mind. his aproach was a socalled "selective optimizing". this system has failed tremendesly in realtime.

    now dont get me wrong i am not saying that your performance cant be reached but you are looking back a certain amount of bars to find the best setting that would have gave you the highest performance during this period, this you do all by refering historical data and thus, you optimize (or adapt as you would like to call it) your strategy parameters to those market circumstaces givin in that period.

    then the next bar or xx bars later you run this same aproach, look back, find parameters, adjust parameters and run next trade.

    i doesnt get more optimized as in your aproach, and what you are doing can be done pretty much in every backtesting software, just not automaticly. that is the only difference imo.

    pls explain what is the difference between you running your aproach or me brute force optimizing a tradestation strategy every bar to find the best parameters ?

    dont get me wrong, i see how your performace is made out of walk-forward testing and i have no doubt your performace is real and *could be* achieved, but i am critizing your *aproach* and find it anything but adaptive.

    i have no idea how one could call your aproach/strategy adaptive


    EDIT: the name was dr clayburg
    http://tinyurl.com/adaptive-or-not (i used tinyurl here because i feel people like him do not deserve free backlinks to promote their useless stuff)

    from his website:
    The Universal System is most effective when the system is selectively
    optimized when necessary. Selective Optimization is a unique application
    of commomly used optimization routines designed to keep the system in
    sync with current market conditions.
     
    #32     Nov 10, 2010
  3. EDIT: i said that wrong, i am not even critizing your aproach, i know this aproach can work, more am i critizing the way you like to name it, which is *adaptive* rather than
    *constant optimized*
    cheers
     
    #33     Nov 10, 2010
  4. jalsck

    jalsck

    I was thinking that the term you used 'bar-after-bar' might have had some meaning that I wasn't familiar with. We are on the same page here.

     
    #34     Nov 10, 2010
  5. nitro

    nitro

    What happens if you detrend the data and feed that to your model? The small edge is probably due to the slight trend in data. I don't see results that break down going long gains and going short gains.

    Also, there is always a hidden curve fitting that people don't take into account - since these sorts of systems appear almost monthly in publications over the years, there is a kind of selection bias that it may have some edge. The only way to quantitatively take that into consideration is to punish a system every time you tried to develop a system on data you have seen or read an article on trading systems.

    The value in a system like this may be as you stated as a system of systems, where this system could protect against strong trends (how did it do in Jan - May of this year and September to today ?), but that is only conjecture on my part since I have not done the study.
     
    #35     Nov 10, 2010
  6. Let us make this a stop & reverse system. You go long when 5 > 30 and short when 30 < 5. No other stops. Simple system.

    For the purpose of being able to compare different approaches I suggest using this data file from 01/2002 for the Qs, daily data timeframe:

    http://www.4shared.com/document/my0i1lo3/QQQQ.html

    Everyone is welcome to compete here!
     
    #36     Nov 10, 2010
  7. jalsck

    jalsck

    I have a price transformation routine setup accumulates the daily natural log return times 100 minus 0.0184. I will give it a spin, it will takes a while to run to completion. The trade report needs a lot of work. Reporting in terms of points is fairly useless over long time periods. I intend to report on Ln returns, remove some less useful stats and add-in CAGR etc. as well as the stats that you mentioned.

    Great point. My personal awareness of hidden curve fitting traps has increased a lot over the last year or so and I am always on the lookout for more!

    I will post the results when the run finishes. I think you are referring to the last one that I posted the details of on this forum, I can re-build any of them though. I was thinking that combining the system with intermediate and long-term systems would work well.

    BTW I don't use any of the systems that I have posted the details of in production. I thought they were interesting systems because of the high dependence on the adaptation process.

    Regards,

    James
     
    #37     Nov 10, 2010
  8. fwiw, i still do not agree with the name you have given the aproach, but the aproach itself i find a wonderfull thing if the calculating does not take to much time. i havent seen any retail platform with something similar yet, if i would want to do this i would need to optimize by hand every new bar to find out the suitable sets of parameter which would have givin me the greatest profit over xx bars for the upcoming trade. i am sure the average retail trader would make great use of it if a platform like tradestation would offer a similar aproach. for this you have my respect
     
    #38     Nov 10, 2010
  9. JackR

    JackR

    Flying:

    See http://www.biocompsystems.com/products/Dakota/

    BioComp products are not widely know but often receive honorable mention in the TASC annual survey (which, in my opinion, is primarily a popularity poll)

    Jack
     
    #39     Nov 10, 2010
  10. jalsck

    jalsck

    For interests sake I have attached a couple of images of the ARM03 run using the detrend S&P 500 data. It's only up to 1987 so far.
     
    #40     Nov 10, 2010