Comparing trading methods

Discussion in 'Strategy Development' started by psheridan050, Apr 13, 2008.

  1. Hi all,

    I’m trying to compare three different trading methods and would like opinions on the pros and cons of such an approach. Each method involves filtering on the daily chart and then implementing a strategy intraday. At this point the filter method and strategies are not so important as determining what benefits or problems are associated with the methods. I’m also curious if others use similar methods for their trading systems.

    Method 1:
    Filter the market using criteria such as daily charts with an RSI(14) > 90. Take all the stocks returned by this filter over a 5 year period and run a simple intraday MA crossover strategy on them.

    Method 2:
    Use the same daily filter RSI(14) > 90 on a basket of 5 stocks over a 5 year period and run the simple intraday MA crossover strategy on them every time they meet the criteria.

    Method 3:
    Use the same daily filter RSI(14) > 90 on a single stock over a 5 year period and run the simple intraday MA crossover strategy on it every time it meets the criteria.
     
  2. To give a little more background as to what I’m asking, I’m wondering if a trading system like method 1 might be more stable because it has a diverse selection of stocks to trade from. I know it would generate more signals, but would this make it more robust? Besides the number of signals, is there any real difference between these trading methods?
     
  3. Vorpal

    Vorpal

    I would suggest you test your unfiltered entry criteria first, then test a second time with your RSI filter to see if it adds any value.

    Test 1: All stocks, no filter, your crossover model

    Test 2: All stocks, RSI Filter, your crossover model.

    If Test 2 doesn't outperform Test 1, then you know your added filter isn't doing much to help.
     
  4. I personally want to see as many signals as possible when I test ideas.

    Sometimes you see something in the data that gives you a clue as to where your ideas may be flawed that you might not see if you only tested one stock or market.

    I don't know what software you use, but I use tradestation.
    The "Pyramid Positions" feature, I feel, is important because it allows you to test what would happen if you placed every trade in one direction (long or short), before a reversal, that your system produced.

    I use this to see if my profit per trade is dramatically different between 1 contract or multiple. If it is, I may have a system that is performing overly well given my date range and the specific market conditions at that time.

    Based on that, I think your systems are essentially the same but just applied over a varied size of data.

    I also agree with Vorpal.
    Best of Luck!!
     
  5. lindq

    lindq


    You need to answer you own question with backtesting software that will let you look at an entire portfolio, and simultaneously backtest on two timeframes, daily and intraday. (Or, a greater number of timeframes if it is useful.)

    You should first begin with the elements of your intraday strategy to see if it is viable at all. (Hint: Very unlikely with MA crosses.)

    Once you find something promising intraday, you then add a daily filter to the portfolio to see if that brings anything to the party.

    In many cases, it will. In addition to daily technical filters, often bringing daily fundamental data into the backtesting can help as well, depending on your strategy. (EPS, PE, etc.)

    If you aren't locked into software, look at Investor RT at www.linnsoft.com. It will do everything I've mentioned above.
     
  6. I think those metods are doomed failure. Because markets have no fraquency characteristics. You should turn you mind in another side of trading. You must find that time then you can get a statistical advantage infront of market.