Help with Monte Carlo Drawdown analysis

Hello,

I was wondering if someone could please do a Monte Carlo analysis for me on the trades in the text file. I've only included the trade results, but if I need to add more I can.

I'm trying to come upt with an alternate, non-simluation based measurement of risk based on risk-of-ruin so I want to compare.

Thanks to anyone who can help.

 

Ooops, forgot the file. Here it is. The actual drawdown was -16689. I think I should probably include the max positive and negative excursions but I've never done a Monte Carlo analysis so I don't know what's needed.

Thanks again.

 

I think a spreadsheet I posted a while ago might do the job.

http://www.elitetrader.com/vb/showthread.php?s=&postid=1246313

The spreadsheet uses a non-parametric bootstrap to give you a distribution of the expected return. For any other statistic than the expected return (e.g. Sharpe, VaR, Risk of Ruin etc.) you need to insert the appropriate formula in cell I2.

Note that if your trade data exhibit serial correlation you can't use this method and need something more advanced.

 

Thanks NoWorries,

I had actually downloaded that before and I copied all the trades in and the formulas all the way down.

But now I have no idea what to do next. It changes all the values randomly, which I assume it's supposed to do. But then I'm lost...

 

what is the statistic you're interested in, and how would you calculate it for the trades you posted? Is it the maximum drawdown, average drawdown, etc.?

 

I'm interested in the theoretical maxdrawdown that happens for 99.9% of situations.

I.e. what the needed capital is for a ROR of less than 0.1%

If you need the max excursions/profits I can repost with all the information from WealthLab...

 

Ok, here's how I would do that: I took your trades and inserted them in column A, I calculated an equity curve in column R, assuming a starting equity from cell Q2 (you can change this). Then I calculated drawdowns in column S (in % of equity), and inserted the formula for the max drawdown (again, in % of equity) in cell I2.
I expanded the number of bootstrap replications from 100 to 1000 in column L. In my case I saw from the frequency distribution in column N and O that for 249 (199+42+7+1) runs, the max dd exceeded 2%, i.e. we can be 75% (1000-249)/1000 sure that the max dd will not exceed 2% ($2,000). Every time you rerun the simulation, these numbers might change a little, so your numbers might be a little different. If you're looking for a higher level of confidence (i.e. 99.9%), you need to ramp up the number of replications (expand the datatable in column L to 10,000). I didn't do this, b/c in the current version it already took a few minutes to recalculate the entire sheet. (NB. make sure automatic recalculation is disabled, i.e. check "manual calculation" under Tools -> Options -> Calculation. Now every time you hit F9, the sheet is recalculated).

NB All this is only based on the drawdown measured at the closing of each trade. During the trade, your drawdown might be larger, for which you would indeed need some data on excursion during the trade. Adding this is not difficult, as it would only require modifying column S

 

NB. A more precise way to estimate the confidence level is to copy and paste (as values) the contents of column L to a new sheet, and sort them in ascending order. If you have 1,000 replications and want to know the upper bound on max drawdown for 90% (900) of the runs, you would just read the value off the 100th line, i.e. 3.6%

 

Thanks for doing this. It does look like I'll have to add the excursion data. The Max DD's are all less then 10k so far but the actual was about 16k.

I have a measure right now that's very close, and gives a value a bit larger than max drawdowns on sims if you limit the number of trades that can be open at a time to 1. It falls apart though when you can have multiple trades open and that's what I'm trying to get to. The measure does use the excursion information so I think I'll definitely need to try it with the spreadsheet you posted.

I'd like to have a predictor of max drawdown based on the actual trade data rather than random samplings so understanding what the actual max drawdowns are can help me figure out what it should be and how to fix the multi-position version.

Any other ideas are welcome as well, but I'll try adding the extra info and post the results.

 

indeed the spreadsheet assumes non of the trades overlaps any other trade. If trades are overlapping it gets more complicated, and drawdowns will be larger for sure. I usually prefer to work with daily (or weekly) returns instead of per-trade returns.