simulation

Discussions about Money Management and Risk Control.
Post Reply
smodato
Senior Member
Senior Member
Posts: 27
Joined: Wed Jul 14, 2004 2:53 am

simulation

Post by smodato » Tue Nov 23, 2004 8:24 am

Hi guys, here:
viewtopic.php?t=139&start=15
I asked some questions about montecarlo simulation, now I would like to bettere define my questions:
if I plan a contract size increase based on a historical list of trades and I grow for example from 10.000 USD to 1.000.000 USD in 3 years, remixing the order of the trades it is obviously impossible to get the same result but how should I judge different outcomes, I mean: if on 10 test I always obtain mucj better results should I be optimistic for the 1 million tested on the series or should I fear a random order of the trades would dramatically change the scenario?
What are you looking at when testing with montecarlo? Could it be reasonable to expect the final result to keep in a rather narrow range (for example 800.000 to 1.200.000)?
Thanks
Bye
Smodato

Hiramhon
Roundtable Fellow
Roundtable Fellow
Posts: 98
Joined: Fri May 09, 2003 12:45 am

Post by Hiramhon » Tue Nov 23, 2004 9:01 am

Please post some of your actual simulation results and a description of the experiments you performed that led to those results.

smodato
Senior Member
Senior Member
Posts: 27
Joined: Wed Jul 14, 2004 2:53 am

Post by smodato » Tue Nov 23, 2004 9:22 am

I explain:
I have 3 strategies and they work together, all together in 6 years of backtesting they gave about 4600 trades, I put all the trades in chronological order intoexcel and applied a money management strategy to start from 80k and risk a fraction of 4% up to 200k, then 3% to 400k and at last 2%, as I told you in 6 years I obtained the virtual equity line I attach hereunder.
Now I would like to investigate what would happen if the same trades occured but in different order, so I would like to "remix" the order of the trades in different random manners and look at the resulta applying the same money management criteria.

I hope I was clear, please ask me if not.
Bye
thanks
Smodato
Attachments
equity.gif
equity.gif (24.09 KiB) Viewed 4478 times

smodato
Senior Member
Senior Member
Posts: 27
Joined: Wed Jul 14, 2004 2:53 am

Post by smodato » Wed Nov 24, 2004 6:51 am

I update you with latest outcomes:
Together with the money management strategy I'm also testing different ways to stop trading a strategy when it's going worse.
So I tested a stop in my strategy once the equity dropped under the 30 periods average, the final results were much worse then the original even though the equity showed a sharper profile and more acceptable drawdowns, another test was performed stopping trading after breakout of the lowest point in last 30 trades and starting again afterbreakout of the highest point in last 30 trades (obviously mesaured on the total 1 contract equity).
These results were more or less the same as stopping when under the 30 periods average.
Last test was to stop at 30 periods low and restart when above 30 periods average. Here the results improved compared to the previous two approaches but were still much worse in final profit than the totally traded strategy.
Obviously this is due to the positive history of the strategy and I'm really willing to find a way to avoid being in a dying strategy so I would accept to drop some trades (the equity i posted here above was actually the composite of 3 strategies on each of whose I dropped trades under 30 periods low).
So do you have any suggestions based on the above?
Further to this I finally managed to "remix" the trades, simply putting in a column random() and ordering the trades according to the random numbers, I did this only on one of the 3 strategies (the one with most trades) and I noticed that in some cases the final outcome was pretty different, mostly around 1.000.000 it happende even to see "only" 300.000 as final equity.
How would you judge this result and the test itself maybe)?
Bye, thanks
Smodato

Post Reply