Simple Moving Average Crossover Strategy

Forums ProRealTime English forum ProOrder support Simple Moving Average Crossover Strategy

  • This topic has 95 replies, 12 voices, and was last updated 1 year ago by avatarCRISRJ.
Viewing 15 posts - 31 through 45 (of 96 total)
  • #190532

    NAS-5m-MACross-v4.1-L1.itf contains 216 lines of code.

    NAS 5m MACross v4.1 L contains 288 lines of code.

    Maybe there is good reason?

    1 user thanked author for this post.
    #190534

    Oh Yes There is a very good reason :’D , i did a mistake by replacing a lot of lines by what @nonetheless have gave me. Thanks for the notice.
    Also i did remove the MM code.
    If i correct that, the code will be : ( no spread, MM off, 2000 capital)

    #190581

    In my opinion, it´s sensless to do a backtest without spread, the results are falsified.

    1 user thanked author for this post.
    #190582

    I agree with you, it’s just to help to find what’s wrong with my backtest wich is 55% less good than Nonetheless Backtest :’D

    1 user thanked author for this post.
    #190583

    I though that, but surely the spread is in the code somehow / somewhere??

    I just glanced over the code, but spread didn’t jump out at me.

    1 user thanked author for this post.
    #190584

    Attached are the results with a constant exposure value of €10k, gives a better visual on the histogram and the curve.

    I am comparing my backtest with the backtest of this “constant exposure” that i don’t know what it is. Thanks

    #190592
    JS

    Line 223 STDDEV = STD[bollperiod]

    STDDEV = STD[bollperiod] (price)

    Missing the (price) ?

    1 user thanked author for this post.
    #190596

    comparing my backtest with the backtest of this “constant exposure”

    this means positionsize = 10000/close

    shows performance if you had risked a constant value (€10k) rather than a constant size per point

    this would definitely explain the difference you are seeing!

    1 user thanked author for this post.
    #190597

    Thank you very much @nonetheless, I understand now the difference !

    #190664

    @nonetheless

    What are your experiences with this type of optimization? I see a lot of small wins and a big SL. In the report, the average gains are also much smaller than the average losses. In my experience, with such values, the strategy usually collapses in sideways phases, so it is not robust. I optimize as much as possible in such a way that the average wins are always greater than the average losses. So you still have the chance to come out plus/minus 0 in sideways phases. Hit rate is then of course lower, between 40-60%.

    how do you see it?

    #190670

    I have some systems with only 55% win rate, but very high risk/reward, others with over 90% wins but risk/reward around .3 or so. Then I have one with the best of both worlds … but very few trades, maybe one per month. There’s always going to be a trade-off between these factors and I think the difference is largely psychological. Either can turn a profit in the long term, it depends on what you’re comfortable with.

    A broker I had years ago liked to say, “No one ever went bankrupt by booking small gains” … and I guess that message stuck as I tend towards high win rate, even if the wins are small. This often means a low breakeven level and high SL that others may find makes them anxious. Personally it doesn’t bother me to carry a trade the wrong way for 1.6 or 1.8% knowing that 90% of the time it will turn around and close for a profit.

    Attached is the first few weeks OOS for v4.1L, positionsize = 1

    (nicely demonstrates one good thing about long-only systems: what you lose on the way down you will probably get back on the way up)

    1 user thanked author for this post.
    #190672

    @nonetheless

    Thanks for the information. I’ll just try it on one of my systems to achieve a high hit rate. Then I’ll try running it next to the original and compare.

    #190777

    Here is my humble contribution to an MA Cross system. I use SP500 here because most of my systems are based on it (many systems, little margin, little spread, portfolio building).
    I solved it completely conventionally, with simple SMAs. The all-important trend seems to be the weekly trend. M15/SMA480 = D1/SMA5

    Here is the system, long/short combined. Maybe there is still room for improvement?


    @nonetheless

    You always have a golden touch. Maybe you can take a look?

    2 users thanked author for this post.
    #190795

    This is the same version (rewritten)of Phoentzs  [ T4 -Multi -SP500 -M5-V3] , with Roberto’trailing. The performance up to 200k is slighthly better, up to 100k is better (see image).

     

    2 users thanked author for this post.
    #190798

    Thank you Mauro, did you really just take another trailing stop? Didn’t think it would make such a big difference.

Viewing 15 posts - 31 through 45 (of 96 total)

Create your free account now and post your request to benefit from the help of the community
Register or Login