Making an almost profitable system profitable

Discussion in 'Trading' started by 1a2b3cppp, Apr 1, 2013.

  1. As you know from my profitable journal thread(s) I currently have a system that I'm using but I'm always interested in other systems that I think may bring something to the table (or simplify something).

    I recently read this thread by IronFist from 5 years ago. Basically he was going long when price closed above the 60EMA and short (or reversing) when it closed below the 60EMA. Special note: let me point out here that what you have to do is open your position on the first tick of the first bar after the close of the bar that closes above or below the EMA. The reason for this is because until that bar closes, you don't know for sure whether it's going to close above or below the MA (failing to take this into account is a common tactic used by snakeoil trading gurus, but that's an entirely different topic).


    His question was on when you should decide to take profits. His observation was that sometimes he had huge winners and other times he would get chopped out, and if there was a way to maximize profitability from that.

    Assume a relative slow MA period (such as the 60 EMA IronFist was using). Begin with a starting condition of going long when price closes above the MA and short when price closes below the MA (and, unless you are using some weird MA, this will correspond to the slope of the MA as well). Sometimes you will have huge soaring winners, but other times you will be chopped to death.

    This in and of itself is not profitable in the long run.

    How can we change this system to make it profitable?

    Time of day? Extra screening criteria (so that not every pass through the MA results in a trade)? Stop order requirements? Open trade management rules?

    Those of you who follow my journal and associated threads know I don't believe in trend following or price predicting, but I'm always willing to have intelligent discussions about it.

    IronFist doesn't post here much recently but I'll send him a PM to see if he wants to join in this discussion.
  2. Lower commissions/spreads/costs ?
  3. 1. You wish to discuss something you don't believe in, therefore the oddz favor you rejecting anything we may present in favor of the ideas in question.....

    2. You do seem to believe in averaging into a trend, based on your journal.

    3. Reading between your lines, it's obvious that you want to use the 60 EMA as your trend direction indicator, averaging into a new signal, taking profits by scaling out as the ema turns over......

  4. Consider this thread entirely separate from my journal thread.

    I average in against the trend.

    You've added some extra rules there. I never said anything about averaging in or scaling out.

    I'm not even saying we having to use a 60 period EMA, I was just mentioning that as a starting point.
  5. Thanks for the PM.

    I remember that thread. I was never able to turn it into a consistently profitable system despite spending some time trying it. I believe slower moving averages have a better potential for something like this. 60 is the fastest I would use. I played around before with some even slower ones like 300 period weighted moving average which seemed to result in even bigger wins but also a lot of back and forth chop when prices goes horizontally. One bar will close above (signal to go long) and then the next will close below (signal to go short and previous long position stopped out for a loss) and repeat.

    I decided that without knowing ahead of time if price was going to chop, this system would be random if it worked or not.

    And if you knew ahead of time if price was going to chop, then any trend following system would work.

    I played with various parameters including some position sizing algorithms, etc., but wasn't making much progress so I stopped.

    Eventually I came up with a version of jjrvat's system from his daytrading 2.0 thread. I came up with concrete rules since to me his thread was more conceptual rather than specific. That system was profitable for the few months I tried it but I stopped because I wasn't confident with it. It may be worth revisiting that one, too. It was basically a 2 MA system with specific price action rules to prevent entry in certain cases.

    Regarding the 60 EMA system, I'm open to discussion about how to make it work. It does produce huge wins in certain cases so maybe minimizing/preventing losses is the way to go.

    I feel fairly confident saying that the solution does not involve adding another trend following indicator.

    Does anyone have the ability to backtest this with specific rules on index future data? I think that would be a good starting point.
  6. dom993


    I did experiment a while back using a pretty long MA (300 to 1000 bars) on a 30min chart ... I was actually using 2 MAs, one from the Highs, one from the Lows, to form a band to help reduce whipsaws. The system was always in, either long or short, no stop nor target. The results at 1000 were positive on all markets that I tried (ES / CL / 6E each 10 years), although low P/F (1.05 to 1.15 if I remember correctly) and pretty large DDs.
  7. kroponer


    Don't have any evidence to support, but most guys that have been profitable on ET have said high VIX is key to profitability, so try backtesting with VIX over 20. Obviously, we would be talking about swing trading over days/weeks and only a couple times a year.

    If you are talking about intraday, no ideas there.
  8. I'm kind of interested in the idea of using longer timeframes like that because I believe that unless you have a specific reason to be trading shorter timeframes, longer timeframes are better.

    Look at a chart of someone going in and out during the waves of an uptrend, for example. Why not just ride the entire thing up?

    Of course, it may be that certain styles only work on shorter frames rather than large frames.

    But really, put a 300 period MA on a chart, color the chart based on slope of the MA, and look at the huge (and profitable) swings.

    Then look at all the times you get chopped out.


    Did you forward test, manually backtest, or automated backtest?

    You had one MA based on each candle's high and one based on each candle's low, correct? How did that help you avoid whipsaws? I ask because I've tried similar stuff in the past with no luck. For an exit did you have to cross the opposite MA? For example, if you are long, is the signal to reverse to a short position that price has to close below the MA formed by the lows of the candles?

    Were your large DDs caused by chop? Seems to me that this type of system would never really have a position go against you for very long due to how MAs work (eg. if price goes against you too far it would close on the opposite side of the MA and reverse your position).
  9. dom993


    1. Automated backtest in Ninja, MKT order at close of 30min bar, 1-tick slippage on entry / same on exit

    2. That helps a lot avoid whipsaws at night, where 30min bars are a lot smaller than the channel created by the 2 MAs. Long closing above the High MA, Short closing under the Low MA.

    3. You are right this system is never on the wrong side for long, DD created by whipsaws in the daytime, and/or large 30min bars (triggered by news events, mostly) triggering a reversal of position far from the MA. That particular aspect might be reduced by using 30min bar boundaries at xx:05 / xx:35 instead of round hour & half hour - I didn't try that though.
  10. dom993


    I just exhumed that strat & tested it again ... results in a nutshell:

    6E: P&L +117k / #trades 553 / DD -20k / P/F 1.51
    CL: P&L +109k / #trades 670 / DD -33k / P/F 1.29
    ES: P&L +24k / # trades 621 / DD -14k / P/F 1.16

    Note the the MA is a 1000-EMA of the prior 2 bars High / prior 2 bars Low
    #10     Apr 2, 2013