Nicola's ProOrder Breakout on any Index

Forums ProRealTime English forum General trading discussions Nicola's ProOrder Breakout on any Index

Viewing 11 posts - 1 through 11 (of 11 total)
  • #8965

    Hi, I have done a small change to the famous TS created by Nicolas posted here:

    http://www.prorealcode.com/blog/automated-breakout-trading-strategy-french-cac40/

    I have removed the MM for the moment just to compare the results and fixed the variables to a % value and not the fixed one choosen by Nicolas.

    The part of the code that I have changed is this:

     

     

    Here some results run with normal IG market spreads.

    Look what happen with Dax and MM 🙂

     

    For this Ts i Propose a different MM. I do not increase the number of contracts but as soon as the first Ts gains 1000 euro I start it on a second index. When also the second index gain 1000 euro I run it on a third index… etc… Nicolas, what do you think about this idea?

     

    Here the whole code:

     

     

    David

    1 user thanked author for this post.
    #8969

    Hi David, thanks for attributing me this nice trading strategy… but it is not mine 🙂 This strategy is an example of ProOrder code and its part of the official documentation.

    Anyway, I don’t understand the factors you set on the first part of  the code? (Max and Min amplitude, etc..) How do you set them? (from line 28 to 30)

    About launch multiple instance of the same strategy on different index, well.. why not!

     

    #8970

    They are choosen from the original.

    In the CAC you have maxAmplitude of 58. This mean if the CAC index is around 4400 you divide 58 by 4400 and you get 0.01318 (%).

     

     

     

    #8971

    Following closer 🙂

    Thanks!

    #8972

    Oh ok! so this is an universal factor made of the amplitude 58 from the CAC40 strategy. That’s clever to have it done this way. But it involves too that all instrument have the same behaviour as the CAC40 index, so maybe you should find an Amplitude that suit well for all instrument instead. That’s the beginning of something..

    #9011

    I Nicolas, here the main problem…

    I could easily find the max amplitude that fit better for all… but… it will be curve fitted!

    How could I find it to be sure that it is not curve fitted?

    I could do an easy forward test:

    • I find the best values for only one index (you already did it. It is 58 for the CAC : -) )
    • I test in on other periods (instead of periods I change index)
    • It gives good performance also on other indexes.
    • The value 58 it is not curve fitted!!

    why should I now make the mistake to find the perfect one on the past that could not work in the future?

     

    Second, can someone that has 200.000 bars test this TS? It would be nice to have it for a bigger period.

    Really easy test… take my code.

    for Dax and cac put spread 1 and 2 contracts

    for Mib put 1 contract and 8 spread

    for eurostocks put 1,5 spread and 3 contracts

     

     

     

    #9013

    You can try to find the mean best amplitude value for all indexes instead of the very better one for all of them. It would be less over-fitted.

    #9071

    It would be curve fitted on 5 curves… don’t you think so?

    #9085

    backtest 200 000 15mn with constant trade volume over the time

    dax=2 contrats spread=1

    cac=2 contrats spread=1

    eur stock=3 spread=1.5

    mib=1 spread=8

    #9088

    Bye!

    #9096

    David, trying to avoid over-fit is very nice but keep in mind that everything in life is over-fit .. from the resistance of your car’s tires to the length of spaghetti..  Human tied to make things better in the future from what they learned from the past and maybe we have to deal with it in trading and maybe over-fit is a part of what would be automated trading.

    I believe that there is no good or bad ways to trade, I used to say that the best strategy is the one that make you a profitable trader.. therefore, apart of learning purpose and understanding, the crusade against over-fitting is a dead end, IMO .. To be clear: don’t over-fit but don’t spend too much time to avoid over-fit. 

    We only have 3 sets of datas to make proof of concept for any trading strategy : time and price (and volumes .. but that’s not the case here). So, somehow, every equity curve of any trading strategy are curve fitted in various ways. Remember I wrote somewhere in our past exchange about this subject, that the first bias we introduce that over-fit our trading strategy concept is: make gain, no loss.

     

Viewing 11 posts - 1 through 11 (of 11 total)

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