looking for strategies to backtest/automate

Discussion in 'Automated Trading' started by travis, Dec 5, 2010.

  1. travis


    If there's any "newbies", or anyone who has a strategy or a vague idea for a strategy and doesn't know how to backtest it and automate it, I've gotten quite good/fast at this, so I would be interested in helping out, and at the same time I will benefit from the new trading ideas, regardless of whether they're profitable or not (I can always modify them).

    I use tradestation for back-testing and IB TWS for automation (through excel). The strategy needs to be quite simple or I won't be able to back-test it, let alone automate it on excel.
  2. bigpapi


    great i found this one recently in an old book, i'll post the name of the book if it works to save the author some embarrassment

    enter long on first 30 min candle that closes above the first 30 minute candle of the day, and market on close.

    do the reverse, go short, for the first 30 minute candle that closes below the first 30 minute candle of the day (and market on close)

    the only thing with this system is that you have to reverse the position from long to short and vice versa if at any time during the day the stock reverses and goes the other way, for example:

    if you entered long because a 30 min candle closed above the first 30 min candle and now you have a candle that closed below the first 30 min candle you have to go short.

    i tried this a few times last week with real money and with a lot of tweaks of my own it has been working well, i did it most recently with HAL NFLX and BIDU, this system completely blows with NKE and FDX
  3. Sell on close Friday. Buy on open Monday.
  4. travis


    Runningbear's strategy is exactly what I needed. Most of my strategies are just like that, with time entries and exits (and other filters). I will test it and get back here to tell you how well it went. Of course for more details such as reports and automation, we need to use private messages. I will automate it for the author (if he wants it) and not for the whole forum. But since you posted a public message, I will post a public reply to say if it works.

    As far as the other strategy I need to ask a few questions, such as when the market opens (is this a strategy for a future or for a stock?).

    I will first test the easiest one, Runningbear's, and then write bigpapi a private message with those questions as I go along in the testing.
  5. This might not be applicable at all, but I don't really trade using a specific strategy. It's somewhat discretionary and I'm more comfortable with my entries, but am always questioning how optimal my hold times and exits are. Would you be able to take a series of trades that were made in the past year and backtest them against tick-level data and determine the optimal hold time?

    Is this something that can be done with Tradestation?
  6. I thought he was just kidding...)
  7. travis


    It would not be a complete strategy if I don't have a univocal method for entries, which I suppose you are unable to provide (you said entries are "discretionary"). One could always do a random entry and tell you what the best exit method is, but that would be beyond my original objective. However we could discuss this in the future: if I ever run out of things to test, I could try to help you with this problem.
  8. travis


    Oh, yeah? Maybe not. It is a good idea to test, whether he was kidding or not. And it would be good to test the opposite, and it would be good to test if, since this might be a tendency, some markets go down in the last few hours of trading on Fridays.
  9. bigpapi


    The stratergy is for equities, but results using the emini should be interesting.

    Personaly I daytrade different equities daily, but if you need particular stocks try HAL NFLX and BIDU

  10. travis


    Ok, I am all done with testing Runningbear's idea. Sorry but it doesn't work. On the other hand, while testing that concept, I came up with a totally different idea for a system and that one worked. I'll have to test it on other futures to see if it works, too. If it does, then I'll have created another few good systems. I still haven't checked the out-of-sample though.

    I am not sure I'll be able to code bigpapi's system, because I am not good at either coding the stop and reverse mechanism and even less at automating it, so it would not be a good idea to back-test it. I've gotten started testing some parts of the system, but I don't think it's enough to be able to say whether it works or not. I'll write you a private message with my contacts so we could maybe discuss it further in a chat (this or other systems).
    #10     Dec 6, 2010