[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Metastockusers] System Tester Results Three card Monte



PureBytes Links

Trading Reference Links


Zen, Monte Carlo simulation is a good adjunct to systems testing. I
use it occasionally. 

There are two main problems with Monte Carlo simulation. The results
of a Monte Carlo simulation are only as good as the quality of the
data that goes into it and the particular simulator design. In my
experience, I've seen many many problems with both of those.

The second issue is the users understanding of probability. For most
people probability is one of the least understood types of math.
Especially the inference. 

If a Monte Carlo simulation says that 90% of the time the tested
system is going to have a positive equity curve over any ten year
period, nothing in the Monte Carlo simulation says that out of the
next one hundred years, the first ten years you trade the system won't
be the ten years in which the equity curve is negative. According to
the laws of probability you could have the next 3 ten year periods
with negative equity curves occur right in a row, and then have
positive equity curves for the next 30 ten year periods after that. Of
course anyone trading such a system would have gone broke long before
the positive equity curve years kicked in.

Their are a whole host of implications both positive and negative
about using MC simulations. Most users think MC simulation is simply
plug and play. They have no idea of how it works or how the results
should be analyzed. 

It's a great tool in very educated hands. 

Thanks for providing some code that let's people give it a try. Very
generous of you. 







--- In Metastockusers@xxxxxxxxxxxxxxx, "zentrader22"
<cris.volker@xxxx> wrote:
> 
> ...and you can add a little bit "Monte Carlo Simulation" to your 
> testing process:
> 
> http://www.zentrader.de/mcsprocess_e.pdf
> 
> First, you can add "data simulation":
> If your system has good system report results not only with your 
> original data, but also with simulated data (which has some relation 
> or properties of the original data), the system might be more stable.
> 
> Second, you can add "system simulation":
> If the Monte carlo simulation shows satisfying profits and 
> low "account drawdowns" the system might be even a little bit more 
> stable and you can think about to trade it.
> 
> Only my two cents.
> 
> I've put the freeware version of my product "Zen Monte Carlo 
> Simulator v3" in the Files section of this Forum. You can download 
> it also from my website:
> http://www.zentrader.de/download.html
> 
> ...and if you think, this idee is useful, you can also buy a 
> professional version:
> http://www.zentrader.de/zen_mcs_english.html
> :-)
> 
> But pay attention:
> These are (in my opinion useful) tools to decrease your risks and 
> improve the quality of your system development process. But there's 
> always a risk concerning trading, which cannot be avoided thru any 
> system or method - that's life!
> 
> bye,
> Volker
> 
> www.zentrader.de
> mail to: info@xxxx
> 
> 
> 
> 
> --- In Metastockusers@xxxxxxxxxxxxxxx, "superfragalist" 
> <jackolso@xxxx> wrote:
> > 
> > Okay, Scott, If you take a look at Quantitative Trading Strategies 
> by
> > Lars Kestner, you see many many tests of various indicators and
> > trading systems from Lars and others. He provides charts showing 
> how
> > each system performed over a ten year period. His charts illustrate
> > the problem with relying on TA for your trading. You will see that 
> he
> > could not find a stock trading system that showed any consistency
> > across more than a couple of years. In addition, if you 
> historically
> > test a system, the historical results have nothing to do with what 
> is
> > going to happen next year. There is no predictabilty value except 
> over
> > very long time frames. 
> > 
> > Your test system could show you would have made 5000% last year, 
> and
> > then when you trade it this year you could lose 5000% this year.
> > 
> > Assuming that back testing is relative to future performance, if 
> you
> > back test your system year by year, instead of over many years at a
> > time, you can put plot the performance of each year and you'll see 
> the
> > risk of trading the system more clearly.
> > 
> > If you want to make money every year, you have to either develop a
> > strategy that's not based on a system of indicator but rather on a 
> set
> > of market conditions, develop a strategy which is not dependent on 
> TA
> > alone, or develop a strategy that employs multiple systems
> > simultaneously across multiple markets. However, if you use the 
> last
> > strategy your overall returns, while possibly positve, will be 
> diluted. 
> > 
> > Most traders learn all of this the hard way. 
> > 
> > Based on what I've said, throwing out last year's losers may be
> > throwing out this year's winners. 
> > 
> > Just because a system failed to work on a stock the last three 
> years
> > doesn't mean it won't work on a stock the next three years. 
> > 
> > If you just want to create a file with the magical results in it, 
> you
> > sort the systems test results, highlight the ones you like and copy
> > them to the clipboard, open excel, paste them in, copy the column 
> with
> > the symbols in it, open word, paste the symbols in as unformatted
> > text, use the replace function to replace the paragaph marks with 
> semi
> > colons and then copy and paste that list into MS to create a 
> history
> > file. Download the historical data and now you're ready to make 
> your
> > fortune. It only takes about 10 minutes. 
> > 
> > Good luck!
> > 
> > 
> > 
> > --- In Metastockusers@xxxxxxxxxxxxxxx, "Scooter" <mariani@xxxx> 
> wrote:
> > > Super,
> > > I did read Mr. Wright's series of articles as you suggested
> > previously. I
> > > found it very informative. I found that I was already doing many 
> of the
> > > things he suggested in developing a system. I have applied many 
> of
> > the ideas
> > > he talked about since reading the articles, more like a book!
> > > 
> > > He does not address the validity of throwing out losers, based 
> on a
> > system
> > > test, from an available list of potential stocks to trade.
> > > 
> > > I know there are a cagillion system developing books to read. I 
> will
> > pick up
> > > a few to read when I have the time.
> > > 
> > > Can you C&P system test results from MS8.0?
> > > 
> > > Scott
> > > 
> > > 
> > >  -----Original Message-----
> > > From: superfragalist [mailto:jackolso@x...]
> > > Sent: Wednesday, April 13, 2005 8:34 PM
> > > To: Metastockusers@xxxxxxxxxxxxxxx
> > > Subject: [Metastockusers] Re: System Tester Results (Redu)
> > > 
> > > 
> > > 
> > >   Looks like you're using the systems tester without much 
> knowledge of
> > >   systems development.
> > > 
> > >   Read a few books on system design and then you'll see you're 
> wasting
> > >   your time.
> > > 
> > >   There's a very good article on this coming up in the May issue 
> of
> > >   Roy's newsletter.
> > > 
> > >   Don't ask for book recommendations because one book won't do 
> it.
> > >   You'll need to read around a dozen or so systems development 
> books to
> > >   get a really good take on how things work. Once you've done 
> that,
> > >   you'll soon figure out that using TA by itself is not the best 
> method
> > >   for picking stocks to trade, the performance of any system 
> based on TA
> > >   is going to dramatically change each year with a lot of losing 
> years,
> > >   there are no indicators that work even reasonably well across a
> > >   variety of stocks or in changing market conditions, and 
> searching
> > >   thousands of stocks every night for something to trade is in 
> the long
> > >   run a losing idea.
> > > 
> > >   Does that mean you can't make money using TA. You can, but you 
> have to
> > >   have a well thought out strategy and apply the strategy 
> appropriately
> > >   for the market conditions.
> > > 
> > >   I'm amazed at how many people that refuse to spend a dime on
> > >   education, information, good data or anything else that might 
> help
> > >   them, but they'll pay a guru thousands for useless trading 
> methods or
> > >   lose thousands more making bad trades. It's baffaling.
> > > 
> > >   You can read Charlie Wright's series of articles called 
> Trading as a
> > >   Business. That will give you some basics about how to approach 
> this.
> > >   Read all the articles in the series. They all have something 
> important
> > >   to say.
> > > 
> > >   http://www.elitetrader.com/tr/index.cfm?s=17
> > > 
> > >   Give Roy's newsletter a try. It's all about MS with the code 
> for
> > >   everything discussed in it. You'll learn more from that in a 
> few hours
> > >   of reading the back issues than you will from playing with the 
> systems
> > >   tester for next few years.
> > > 
> > >   www.metastocktips.co.nz
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > >   --- In Metastockusers@xxxxxxxxxxxxxxx, "Scooter" 
> <mariani@xxxx> wrote:
> > >   > O.k.,
> > >   > I have finally cooked the indicators enough (nothing new) to 
> come up
> > >   with a
> > >   > system that is profitable over several different period back 
> tests,
> > >   at least
> > >   > in MS. My question is, do I remove the loosing or zero trade 
> tickers
> > >   from my
> > >   > potential list of tradable tickers I will run my Exploration 
> on when
> > >   looking
> > >   > for a new trading opportunity? MS seems to be keen 
> on "cleaning
> > up" by
> > >   > removing the losers. Doesn't make sense to me to say because 
> the
> > >   tickers did
> > >   > not hit or make a profit over x amount of periods they will 
> not
> > make a
> > >   > profit in the future. Or does it increase my potential
> > profitability by
> > >   > removing them? Can I somehow export the remaining ticker 
> symbols to
> > >   excel or
> > >   > csv so I can adjust my HSQuote list accordingly?
> > >   > Thanks, Scott
> > > 
> > > 
> > > 
> > > 
> > > 
> > >
> > -------------------------------------------------------------------
> ---------
> > > --
> > >   Yahoo! Groups Links
> > > 
> > >     a.. To visit your group on the web, go to:
> > >     http://groups.yahoo.com/group/Metastockusers/
> > > 
> > >     b.. To unsubscribe from this group, send an email to:
> > >     Metastockusers-unsubscribe@xxxxxxxxxxxxxxx
> > > 
> > >     c.. Your use of Yahoo! Groups is subject to the Yahoo! Terms 
> of
> > Service.





------------------------ Yahoo! Groups Sponsor --------------------~--> 
In low income neighborhoods, 84% do not own computers.
At Network for Good, help bridge the Digital Divide!
http://us.click.yahoo.com/EpW3eD/3MnJAA/cosFAA/zMEolB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/Metastockusers/

<*> To unsubscribe from this group, send an email to:
    Metastockusers-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/