Monte Carlo Analysis

Forums › ProRealTime English forum › ProOrder support › Monte Carlo Analysis

Viewing 15 posts - 1 through 15 (of 39 total)
1 2 3
  • 08/20/2016 at 1:52 PM #11939
    GraHal

    Anybody know of the best / easiest / cheapest / most reliable results (any / all of these! ha)  commercial software to do Monte Carlo analysis of a Bot generated / backtested on PRT please?

    It be great if with a few clicks of a mouse the results of a PRT backtest could be ‘flipped’ into a ‘monte carlo tool’ then press go and wait for the results?? 🙂

    I’m sure above is a utopian dream of what is available, but how close can we get?

    All comments / questions appreciated.

    Thank You

    GraHal

    08/20/2016 at 2:01 PM #11940
    Cosmic1

    Was thinking the exact same thing last night. Subscribed!

    09/03/2016 at 10:24 AM #12627
    GraHal

    Hi @Cosmic

    Follow the link below for a simple to operate tool (a few mouse clicks) that does give more confidence in a Bot.

    Basis of operation is that, over (for example 10,000 bars) the same / similar trades are likely to occur in the future … but NOT in the same order. FreeMC mixes up the Gain / Loss results 2,500 times for a given starting equity value.  So you can see if a particular Bot is likely to ‘blow up your acount’ and  also what are the median drawdowns and median returns.

    You can ‘drag’ the results of a Bot backtest into excel and then use the ‘Gains’ column to drive the freemc.xls below. Let me know if you (or anybody) gets stuck. Full instructions are on the 2nd sheet of the freemc.xls.

    Its not strictly Monte Carlo Analysis … I think it’s nearer to ‘Bootstrap’?

    http://www.kjtradingsystems.com/freemc.xls

    I am advised by KJ that the ‘more complicated, feature packed Monte Carlo spreadsheet’ is not available to buy any longer, but is still given free to students of his course.

    KJ does say that the FreeMC vb code is not locked and so can be amended / improved! Any Excel vb coders fancy trying adding a few bells and whistles?? Nicolas maybe (as if you aren’t snowed under with work already)?

    If anybody has anything better let me know please?

    Enjoy & Prosper

    GraHal

    2 users thanked author for this post.
    avatar jonjon, avatar Nicolas
    09/04/2016 at 9:10 AM #12646
    quo

    Hi Grahal,

    Have you try Equity Monaco? its for me the simplest and faster way to do it

    you can downloaded here, its free:

    http://www.tickquest.com/?page_id=70

    if you want to see how tu use it

    https://www.youtube.com/watch?v=GseuHy5JD1A (sorry its in spanish! 🙂

    hope it helps.

     

     

    4 users thanked author for this post.
    avatar AsgerK, avatar jonjon, avatar GraHal, avatar Nicolas
    09/05/2016 at 10:33 AM #12686
    GraHal

    Hey Quo that is great, I will try it out for sure.

    The user manual / .pdf (in english) gives a straightforward intro to monte carlo that ayone can understand.

    The .pdf also helps readers to apprecaiate that Monte Carlo testing on a Bot before going live and risking real ££££s is essential to reduce the chance of a wipe-out of capital and so maybe ‘out of the game’!

    Many Thanks

    GraHal

    09/09/2016 at 7:23 PM #12908
    Elsborgtrading

    Hi GraHal. So I’m wondering if I got this right?

    I did a simulation with only 1 position on a USD/JPY fx as the instruction said. I exported the Proorder simulation table -used the”Closed Postion List” and Used the “Abs Perf” column. Since it was in JPY I converted it to USD- removed all the “.” in the numbers so I only had the “,” So far so good, it produces some numbers and graphs. It seems to work.

    But how do we interprets the values?

    There would properly not be an exact number, as it would be a matter of opinion. The best values are similar the utopia equity curve(0%ruin, and infinite Return/DD)

    what I would like to get is yours, and other traders points of views on the values in the MC Simulator

    These were my own result.
    Cheers Kasper

    1 user thanked author for this post.
    avatar jonjon
    09/09/2016 at 8:22 PM #12911
    GraHal

    Hi Kasper

    I’m sure the ‘how to interpret’ is on Sheet 2 or 3 of the FreeMC .xls workbook?

    Main one for me is the ‘Ruin’ column and your Bot shows that you risk an 18% chance of ruin / wipe-out of your account if you were to try for that 509% gain in 1 year! ha Also a 55% chance of ruin if your starting equity is $650 as shown in the 1st row of the yellow table.

    So you need to increase your starting equity until you get a 0% in the ‘Ruin’ (wipe-out) column.

    Even then though, what is the confidence level of the Ruin % figure? The instructions mention 50% … I’ll have to go back and read again as this seems a very low confidence level? Higher confidence will be gained by more itterations / scrambles of the Profit / Loss figures. The FreeMC does 2,500 scrambles of the order in which the profit / Loss might result from a Trade.

    The ‘Median’ columns will be for example … your median return out of 2,500 scrambles of Profit / Loss … but would you want to live with that 18% risk of getting wiped-out and not getting that $11,583 because you started out with only $2,275 in your account?

    Hope that Helps?

    GraHal

     

     

    1 user thanked author for this post.
    avatar jonjon
    09/09/2016 at 8:34 PM #12912
    Nicolas

    Hey I’m a fan of your work, keep going guys 👍

    1 user thanked author for this post.
    avatar Elsborgtrading
    09/09/2016 at 10:16 PM #12914
    GraHal

    Thank You Nicolas 🙂

    I said I’d explain the ‘50%’ … this is the  chance of achieving the Median figures … which of course it would be because Median is the midpoint of a range of values and so there woud be an equal chance of getting a higher or lower result in actual trading results.

    You might think … well what good is that, my profit might be lower! Well yes it might, but there is every likelihood that your profit will be ‘of the order of / similar to’ the median figure.

    The reason for this is because to achieve profit figures ‘loads less or loads more’  than the Median Profit shown in the ‘FreeMC Yellow Table’ will be less and less likely the further the result is from the median figure shown in the Results Table.

    GraHal

    09/10/2016 at 12:39 PM #12917
    Elsborgtrading

    Hi Grahal.

    Yes there is some explanation on 2nd sheet, but nothing really to confirm my search. As you already wrote, it explains the 50% chance for the reason you listed. Anyway I got a more clear information now, because as I understood the yellow box is simulated runs within the first year, and not as I initially thought- runs over several years- one for each row. It makes more sense now. It helps you pick your starting capital.I can also see by increasing my stating capital to, lets say 10000$ it will result in a 0% ruin.

    but let say you implement reinvestment of your earnings- like suggested in 2nd sheet(positionsizing). I’m not sure, but I think the MC analyses would never derive in a 0% ruin at any time. I have a hunch that only investing 1-2% of your earnings- as the rule of thumb in real trading, would keep you on the safe side. But that is not what anyone is trying to achieve in this site(including my self until now). what you think?

    Cheers. Kasper

    09/10/2016 at 6:30 PM #12931
    GraHal

    Hi Kasper

    Yes the FreeMC assumes a 1 contract trade and 1 year worth of Profit / Loss. The Yellow table shows 11 rows of results each with a different starting equity.

    Yes I agree, if you were to input the minimum starting equity for a 0% Ruin based on a 1 Contract Trade and then used that same value of starting equity for a 10 contract Trade together with associated ’10 Contract Profit / Loss’  then the FreeMC would give different Ruin %.

    If you were reinvesting earnings then you would have a higher starting equity anyway. Thing to do would be to ‘trick the FreeMC’ that the 10 Contract Trade Profit / Loss is from a 1 Contract Trade (FreeMC assumes a 1 Contract Trade) but you would need to increase the starting equity by a factor of 10  also to get the same Ruin %.

    I’ve not tried above so I stand to be corrected?

    I don’t bother  making sure my Profit / Loss is for exactly 12 months. If it is 3 years worth of figures then I just divide the Median Profit etc by 3 if I want an annual figure. The Ruin % is valid for those same 3 year Profit / Loss figures.

    Apologies if you think my explanations ‘laboured somewhat’, but I try to benefit others reading also.

    Cheers

    GraHal

    1 user thanked author for this post.
    avatar Bard
    09/11/2016 at 10:15 AM #12935
    Elsborgtrading

    Hi Grahal.

    This is the cool thing about this site. You try out, investigate, discover, and sparring with others. In my MC sim quest I have come to this. Not only do every one need to question their data, we also need to double check. I found out by trying the other program from Tickerquest, that is was similar output to the freemc, but it also has more options and guided information. I also discovered that in my conversion from JPY to USD I made a mistake by a factor 10. very devastation for my MC sim, as it turned out to not be 55% ruin but 1%.

    Happy days!! 😀

    Kasper

    09/11/2016 at 11:33 AM #12939
    Elsborgtrading

    I just found out that exporting data from PRT only export the first 1000 rows, and not the whole trade history. is there a way to change this?

    Cheers Kasper

    09/11/2016 at 5:00 PM #12949
    Elsborgtrading

    Hi Grahal. To answer you question, I think you can “cheat” the MC sim to compensate for the positionsizing. Also to see if your positionsizing are not “curve”fittet or in any way to good to be true. The MC from tickerquest has such an ability but I could not get it to work. I simply don’t produce any output. However I had some problems installing it, so perhaps that’s why.

    Cheers Kasper

    09/12/2016 at 9:48 AM #12969
    GraHal

    Hi Kasper

    Yes all good points you are making, as you say we can progress faster due to each posting our own findings.

    I’ve not tried the TickQuest MC yet, although I have speed read the instructions … it looks good, I will try and do a test very soon to see if I have same problem as you.

    Re export from PRC … do you mean drag from PRC and drop into excel? This is the only way I have found to do it so far.

    Coffee & Toast now … mmm! ha

    GraHal

  • Author
    Posts
Viewing 15 posts - 1 through 15 (of 39 total)
1 2 3

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
technical analysis fair

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