Basic Strategy question

Discussion in 'Strategy Development' started by StrHead, Aug 16, 2009.

  1. StrHead


    1st time posting after watching for a while so excuse me if I am going over old ground here.

    If I have a strategy which, on paper, seems profitable on 1 product (ES to be specific) or perhaps 1 class of products but seems to behave almost exactly the opposite on another class (ETFs such as GLD to be precise again) should I apply the strategy with the confidence the results seem to give on ES or should I look for some meaning in the fact that it does work on other products?

    When I started looking at the basic strategy it seemed to me to be a price action only based one, in which case I am surprised the performance is so markedly different between different products.

    Thanks for your help and entertaining me in the last months!
  2. 1) Apply the strategy to different markets. The appearance that stocks and gold seem to be inversely correlated may only be a temporary phenomenon or something that could persist longer.
    2) I hope you've been "informed" just a little bit from this site. :cool:
  3. Can you be more specific what you mean by "price action" without disclosing any of your important details?
  4. StrHead


    Sure. I have a very basic strategy on ESbasically saying given a candle looking like 'x' go long/short and place stop and limit orders at yesterdays high and low.

    So really just a function of buyers and sellers as far as i thought - hence my surprise at the differing behavior in two products when I always thought there was a lot of speculators in both.

    I guess this is a general question from someone starting out trying to really get a well defined 'strategy' regarding how ideosynchratic (after several spelling attempts I give up on that one! :D )you would expect a simple strategy to be and what pitfalls can befall me at this point.
  5. This is more like micro-pattern trading and I am not sure what the connection is between price action and patterns in general. The problem is that the same price action can give rise to different patterns depending on how one determines them.
  6. StrHead


    i guess my question is less about the type of strategy as I think that would limit what I have the opportunity top learn with this question.

    More, let me ask what can I learn about the validity of a strategy given its performance in backtesting across different products (perhaps two different equity products, perhaps FI and equity etc).

    It just makes me nervous seeing something looking good on ES but bad on everything else - i have some experience with stats etc as well as markets but I am much more interested in what real experience has told people rather than what books say (which is often little).

    If it's relevant I backtest with 2001 - 2005 data in-sample and look at 2005 to now out of sample.
  7. vikana

    vikana Moderator

    In my opinion its perfectly acceptable that a system for ES doesn't do well on e.g. corn or individual stocks. Some systems do, but many don't.

    I would like to see an ES system to OK on NQ and YM, and some of the relevant ETFs.
  8. StrHead


    Ok great.

    I am using some free software ProRealRime for my strategy design. It is by far the best feee thing I have found out there but it only gives me daily data.

    I have tested my strategy in this and it seems in total good though it has some looonnng shallow drawdown periods.

    I would like to see if it works on a shorter timeframe - I wonder if anyone is able to run it for me via something more sophisticated in exchange for me on 4hr/1hr ES to see if it is still valid?

    I would also love trying something I have seen mentioned before but do not have the technology to try myself which is applying a filter based on the VIX. It makes sense to not do this trade (intuitively - have not tested it) with a high VIX so I would love to see what that does to the curve.

    I will now attempt to post in-sample equity curve, out of sample and the total. Apologies if this does not work - i have not tried it before.

    here is in sample with luck....
  9. StrHead


    and outofsample
  10. StrHead


    and the total as it looks nice!

    note that I would do the money management differently here taking size depended on max loss - here i had to use just 1 contract.

    thanks for your help so far!
    #10     Aug 20, 2009