|
PureBytes Links
Trading Reference Links
|
Simply put, markets and market prices are created by people, not by
random number generators. People, unlike RNGs, have both emotions
and memories, especially of the recent past.
This combination leads to a serial correlation effect upon the
trading that people do in that they tend to keep doing what has been
working in the recent past. More and more traders are drawn towards
that sort of trading and markets start to adjust to this.
Eventually, the method breaks down but traders keep trading the same
method simply because it worked in the past. This process, after a
good period, leads to a longer series of losing trades than might be
expected from a random simulation.
Something to think about...
--- In amibroker@xxxxxxxxxxxxxxx, "qweds_560" <qweds_560@xxxx> wrote:
> Hi Quentin,
>
> Think it was called Crystal Ball (some sort of Excel add-in).
>
> I can give you qualitative (as opposed to quantitative) reasons as
> to why there is positive serial correlation between trade results.
>
> Let me know if interested.
>
> Sam
>
> --- In amibroker@xxxxxxxxxxxxxxx, "alonzo798" <qqquentin@xxxx>
wrote:
> > Sam,
> >
> > Your 3 cents are appreciated.
> >
> > I think (but haven't done anything to verify it) that indeed the
> > returns of a trading system will most often show some degree of
> serial
> > correlation. And this weakens somewhat the results of a Monte
> Carlo
> > simulation.
> >
> > The Big Question being : to what extent ?
> >
> > It would be enlightening to play with the MC simulators you've
> heard
> > of and see how the outcomes change as a function of the degree of
> > serial correlation.
> >
> > Do you have a link or a name ?
> >
> > TIA,
> >
> > Quentin
> >
> >
> >
> >
> >
> > --- In amibroker@xxxxxxxxxxxxxxx, "qweds_560" <qweds_560@xxxx>
> wrote:
> > > Sorry guys, I felt like interrupting your discussion with my 3
> cents.
> > >
> > > I think you are on the right track with your caveat "...Monte
> Carlo
> > > simulation supposes that the individual trade returns are
> completely
> > > independent from each other, which may not be the case..."
> > >
> > > If correct, that there is serial correlation between
individual
> > > trade returns, then any stats calculated on the basis of
> > > independency of events don't mean much. With regard to Monte
> Carlo
> > > simulation, I think you will tend to underestimate the risk of
a
> > > series of losing trades.
> > >
> > > I for one believe in the serial correlation of returns,
> especially
> > > those generated from trading systems.
> > >
> > > Regards
> > >
> > > Sam
> > >
> > > P.S. I have heard of some Monte Carlo generators where you can
> add a
> > > degree of serial correlation in.
> > >
> > >
> > >
> > > --- In amibroker@xxxxxxxxxxxxxxx, "alonzo798" <qqquentin@xxxx>
> wrote:
> > > > Marco,
> > > >
> > > > First point :
> > > >
> > > > What I meant is just that the info regarding how to format
the
> data
> > > > output from Amibroker in order to make it usable by Equity
> Monaco
> > > is
> > > > buried in the middle of the EM manual, not at the beginning
of
> it
> > > as
> > > > one would have expected.
> > > >
> > > > I wouldn't worry about the "black box" or reliability issues.
> > > > I am definitely not a programmer, but randomizing a list of
> > > returns,
> > > > calculating an equity line from them and repeating this
> process
> > > many
> > > > times over doesn't look like rocket science programming.
> > > > It looks more like something a layman like me can do with a
> > > > spreadsheet. And this is probably the reason why TickQuest
> gives
> > > away
> > > > Equity Monaco for free.
> > > >
> > > > Second point :
> > > >
> > > > I don't know of a method that would estimate the "error
rate"
> of a
> > > > Monte Carlo simulation, but this certainly doesn't mean that
> such
> > > > a method doesn't exist.
> > > > One approach that comes to mind : let's say you've come up
> with a
> > > > system that has a 50 % success rate in Amibroker. You could
> > > compare
> > > > its sequence of wins/losses with the one from a coin toss.
If
> for
> > > > instance the runs are much shorter with your system, this
> would be
> > > an
> > > > indication that your individual trade returns are
correlated.
> This
> > > > would weaken the conclusions derived from a subsequent Monte
> Carlo
> > > > analysis, since MC would ignore the correlation of returns.
> > > >
> > > > I recall reading something about this on the Web, and will
get
> > > back at
> > > > you if I manage to find it again.
> > > >
> > > > Quentin
> > > >
> > > >
> > > > --- In amibroker@xxxxxxxxxxxxxxx, Khamsina11
<Khamsina11@xxxx>
> > > wrote:
> > > > >
> > > > > Hi Quentin,
> > > > >
> > > > > A very instructive message !
> > > > >
> > > > > At the end of your post, you dealt with 2 drawbacks :
> > > > >
> > > > > 1) The Equity Monaco software does not show how
it "handles"
> the
> > > > trade
> > > > > data fueled by Amibroker.
> > > > >
> > > > > Do you mean that it is a kind of "black box" (or worse it
> > > performs
> > > > > unreliable simulations) ?
> > > > >
> > > > > 2) "....Monte Carlo simulation supposes that the
individual
> > > trade
> > > > > returns are completely independent from each other, which
> may
> > > not be
> > > > the
> > > > > case..."
> > > > >
> > > > > Is there any method to estimate the "error rate" of a MC
> > > simulation
> > > > ?
> > > > >
> > > > > Tia for your help,
> > > > >
> > > > > Regards,
> > > > >
> > > > > Marco
> > > > >
> > > > >
> > > > >
> > > > > alonzo798 a écrit :
> > > > >
> > > > > >Scott,
> > > > > >
> > > > > >Let's say I have built with Amibroker a trading system
that
> > > shows
> > > > > >good stats.
> > > > > >
> > > > > >These stats are a function of how the sequence of the
trades
> > > > produced
> > > > > >by the system took place. For instance, perhaps my
> drawdowns
> > > were
> > > > low
> > > > > >because my worst losers were distant from each other. If
> some
> > > of
> > > > them
> > > > > >had taken place in a row, my worst drawdown could have
been
> > > > > >devastating.
> > > > > >
> > > > > >So, is it a good system or not ? Is it robust, or have I
> just
> > > been
> > > > > >lucky ?
> > > > > >
> > > > > >This is where I use a Monte Carlo simulation. Basically,
> it
> > > puts
> > > > all
> > > > > >the returns of the system's individual trades in a jar
and
> then
> > > > plays
> > > > > >the innocent hand. When the jar is empty, I have another
> equity
> > > > curve
> > > > > >based on the same trades, but arranged differently.
> > > > > >
> > > > > >A simulator like Equity Monaco will do that 10 000 times
in
> less
> > > > than
> > > > > >a minute.
> > > > > >
> > > > > >It will then allow me to answer questions such as : "What
> is the
> > > > > >chance that the system can lose money in a one-year
> period?", or
> > > > "What
> > > > > >is the largest drawdown I can expect with a 95 %
confidence
> > > level?"
> > > > > >not simply based on a single equity curve, but on 10 000
> > > randomized
> > > > > >ones.
> > > > > >
> > > > > >It gives me a better grasp on the robustness of the
system
> and
> > > on
> > > > > >what to expect from it.
> > > > > >
> > > > > >One caveat : such a simulation supposes that the
individual
> > > trade
> > > > > >returns are completely independent from each other, which
> may
> > > not
> > > > be
> > > > > >the case.
> > > > > >
> > > > > >And one last point : the way to correctly "feed" EM with
> > > > > >trade data from Amibroker is hidden in the middle of the
EM
> doc
> > > > (page
> > > > > >19).
> > > > > >
> > > > > >Hope this helps,
> > > > > >
> > > > > >Quentin
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >--- In amibroker@xxxxxxxxxxxxxxx, "Scott Mariani"
> <mariani@xxxx>
> > > > > >wrote:
> > > > > >
> > > > > >
> > > > > >>Quentin,
> > > > > >>
> > > > > >>Thanks for the link. I tried it with my AB output but I
am
> not
> > > > sure
> > > > > >>how to best analyze the results. Would you share your
> approach
> > > to
> > > > > >>using this tool.
> > > > > >>
> > > > > >>Thanks, Scott
> > > > > >>
> > > > > >>
> > > > > >>
> > > > > >>
> > > > > >>
> > > > > >>--- In amibroker@xxxxxxxxxxxxxxx, "alonzo798"
> <qqquentin@xxxx>
> > > > > >>
> > > > > >>
> > > > > >wrote:
> > > > > >
> > > > > >
> > > > > >>>Hi Marco,
> > > > > >>>
> > > > > >>>Do a Google search on "TickQuest Equity Monaco".
> Download,
> > > > > >>>
> > > > > >>>
> > > > > >install
> > > > > >
> > > > > >
> > > > > >>>(it's free).
> > > > > >>>Export/import your Amibroker trade data into EM. It
works
> > > like a
> > > > > >>>charm.
> > > > > >>>
> > > > > >>>Quentin
> > > > > >>>
> > > > > >>>
> > > > > >>>
> > > > > >>>--- In amibroker@xxxxxxxxxxxxxxx, Khamsina11
> > > <Khamsina11@xxxx>
> > > > > >>>
> > > > > >>>
> > > > > >>wrote:
> > > > > >>
> > > > > >>
> > > > > >>>>Hi,
> > > > > >>>>
> > > > > >>>>Is there a Monte Carlo add-on for Amibroker ?
> > > > > >>>>
> > > > > >>>>Thanks in advance for your help,
> > > > > >>>>
> > > > > >>>>Marco
> > > > > >>>>
> > > > > >>>>
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >Please note that this group is for discussion between
users
> > > only.
> > > > > >
> > > > > >To get support from AmiBroker please send an e-mail
> directly to
> > > > > >SUPPORT {at} amibroker.com
> > > > > >
> > > > > >For other support material please check also:
> > > > > >http://www.amibroker.com/support.html
> > > > > >
> > > > > >
> > > > > >Yahoo! Groups Links
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
------------------------ Yahoo! Groups Sponsor --------------------~-->
What would our lives be like without music, dance, and theater?
Donate or volunteer in the arts today at Network for Good!
http://us.click.yahoo.com/Tcy2bD/SOnJAA/cosFAA/GHeqlB/TM
--------------------------------------------------------------------~->
Please note that this group is for discussion between users only.
To get support from AmiBroker please send an e-mail directly to
SUPPORT {at} amibroker.com
For other support material please check also:
http://www.amibroker.com/support.html
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/amibroker/
<*> To unsubscribe from this group, send an email to:
amibroker-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|