RE: [amibroker] Re: Backtest multiple systems across multiple timeframes
Thanks Tomasz,
It s gonna take me some time to understand your idea. I'm not very
satisfied with the overall way AB is handling this kind of topic.. anyways
thanks for the assistance. could you explain a little more on how you put
the #include statements?
How many systems did you plug together? How many different timeframes? Did
optimization for capital allocation work? (optimize percentrisked)
Greets,
Matthias
From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf
Of tf28373
Sent: Donnerstag, 9. September 2010 11:01
To: amibroker@yahoogroups.com
Subject: [amibroker] Re: Backtest multiple systems across multiple
timeframes
Hi Matthias
Some time ago I was working on the same problem. The solution I have come
across is as following:
1) use #include command in the main code
2) inside the #include function do like this (of course all is just an
example which will need adjustment to your needs):
a) function(parameter1, parameter2,...,timeframe,...,parameterN)
{
switch (timeframe)
{
case 60: TimeFrameSet(inHourly); break;
case 15: TimeFrameSet(in15Minute); break;
case 5: TimeFrameSet(in5Minute); break;
default: break;
}
//here comes the calculations of channels, threshold,
averages, oscillator, etc, everything you need to obtain
signals' conditions
TimeFrameRestore();//AGAIN - it is just an example, by the way - sorry for
illegable layout
switch (timeframe)
{
case 60: channel1=TimeFrameExpand(chan1,inHourly,expandLast);
channel2=TimeFrameExpand(chan2,inHourly,expandLast);
oscillator1=TimeFrameExpand(osc1,inHourly,expandLast);
break;
case 15: //the logic of code the same as above...
break;
case 5: //as above break;
default: channel1=chan;
channel2=chan2;
oscillator1=osc2;
break;
}
//here comes the buy/sell/cover/short/stop conditions and
position sizing, etc
}
Although using #include results in slower code exection, it is a kind of
idea to handle different timeframes system backtest, so I hope that even if
it does not help directly, it will at least inspire you to find your own
solution.
Regards
Tomasz
--- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com ,
Matthias meridian...@... wrote:
Hi,
thanks to the contribution of Ed Pottasch, supported by Bruce, I was able
to dig a little deeper into Amibroker coding. Everybody who is interested in
applying multiple systems on the same underlying simultaneously should look
here, great piece of work:
http://finance.groups.yahoo.com/group/AmiBroker-at/message/5349
Thanks Ed, thanks Bruce.
Unfortunately, I stumbled across a couple of questions when backtesting
multiple systems across different timeframes, hope someone can help, sorry
for the post being a bit lenghty.
Both systems are traded on the same underlying, in order to make things
easier for AB (Which is a bit strange) I used the same set of data, just
renamed it. both systems operate on the same timeframe, say 15mins.
Question 1:
I use the same variable percentrisked for both systems. Wanted to
optimize for percent risked (only!, this is NOT shown in the example below),
so to say capital allocated to each system for the smoothest equity curve,
AB keeps crashing... Can I use the same variable name in each sub-section or
are there limits? should I dedicated percentrisked1 to system1 and
percentrisked2 to system2 only? I am not a programmer, but for my
understanding, both variables are local, so AB should not be crashing...?
Is using Setoption in this context appropriate or would it result in
wrong values?
if(Name()==DAX_CFD_day1)
{
percentrisked=2.0;
factor=Optimize(ATR-Factor,8.5,3,12,0.5);
number=(percentrisked)/(ATR(14)*factor)*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
.systemlogic here
}
if(Name()==DAX_CFD_day)
{
percentrisked=Optimize(Bolli,0.6,0.5,1,0.1);
sl=2;//Optimize(sl,2,2,2.5,0.5);//good:6
number=(percentrisked/(Ref(ATR(14),-1)*sl))*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
SetOption(FuturesMode,True);
SetTradeDelays(1,1,1,1);
Equity(1);
... systemlogic here
}
Question 2:
Both systems above use 15min timeframe. Another system is using 1hr
timeframe and is trading FX. I was not able to re-write the logic so that I
could backtest the 3 systems with AA settings 15min timeframe. Any ideas? I
do have about 8 systems, lowest timeframe is 5min, highest timeframe 4hrs.
That would require a lot of re-writing... Am I alone with my I have too
many-systems Problem or am I missing somehting?
original logic in 1hr timeframe:
percentrisked=0.007;
sl=4.5;
tp=2.5;
number=((percentrisked)/(Ref(ATR(14),-0)*sl));
SetPositionSize(number,spsPercentOfEquity);
SetOption
[amibroker] Backtest multiple systems across multiple timeframes
Hi,
thanks to the contribution of Ed Pottasch, supported by Bruce, I was able to
dig a little deeper into Amibroker coding. Everybody who is interested in
applying multiple systems on the same underlying simultaneously should look
here, great piece of work:
http://finance.groups.yahoo.com/group/AmiBroker-at/message/5349
Thanks Ed, thanks Bruce.
Unfortunately, I stumbled across a couple of questions when backtesting
multiple systems across different timeframes, hope someone can help, sorry for
the post being a bit lenghty.
Both systems are traded on the same underlying, in order to make things easier
for AB (Which is a bit strange) I used the same set of data, just renamed it.
both systems operate on the same timeframe, say 15mins.
Question 1:
I use the same variable percentrisked for both systems. Wanted to optimize
for percent risked (only!, this is NOT shown in the example below), so to say
capital allocated to each system for the smoothest equity curve, AB keeps
crashing... Can I use the same variable name in each sub-section or are there
limits? should I dedicated percentrisked1 to system1 and percentrisked2 to
system2 only? I am not a programmer, but for my understanding, both variables
are local, so AB should not be crashing...?
Is using Setoption in this context appropriate or would it result in wrong
values?
if(Name()==DAX_CFD_day1)
{
percentrisked=2.0;
factor=Optimize(ATR-Factor,8.5,3,12,0.5);
number=(percentrisked)/(ATR(14)*factor)*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
.systemlogic here
}
if(Name()==DAX_CFD_day)
{
percentrisked=Optimize(Bolli,0.6,0.5,1,0.1);
sl=2;//Optimize(sl,2,2,2.5,0.5);//good:6
number=(percentrisked/(Ref(ATR(14),-1)*sl))*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
SetOption(FuturesMode,True);
SetTradeDelays(1,1,1,1);
Equity(1);
... systemlogic here
}
Question 2:
Both systems above use 15min timeframe. Another system is using 1hr timeframe
and is trading FX. I was not able to re-write the logic so that I could
backtest the 3 systems with AA settings 15min timeframe. Any ideas? I do have
about 8 systems, lowest timeframe is 5min, highest timeframe 4hrs. That would
require a lot of re-writing... Am I alone with my I have too many-systems
Problem or am I missing somehting?
original logic in 1hr timeframe:
percentrisked=0.007;
sl=4.5;
tp=2.5;
number=((percentrisked)/(Ref(ATR(14),-0)*sl));
SetPositionSize(number,spsPercentOfEquity);
SetOption(maxopenpositions,1);
CCIperiod=Optimize(CCI,36,34,40,1);
CCIthreshold=optimize(CCIthres,89,88,96,1);
MAperiod=Optimize(maperiod,7,6,8,1);
MA1=MA(C,MAperiod);
MA2=MA(Ref(C,-2),MAperiod);
CCIshort=CCI(CCIperiod)=ccithreshold;
CCIbuy= CCI(CCIperiod)=-CCIthreshold;
Buyok=Ref(CCIbuy,-1) AND Cross(MA1,MA2);
Sellok=CCIshort;
Shortok=Ref(CCIshort,-1) AND Cross(MA2,MA1);
Coverok=CCIbuy;
timestart=02;
window=17
Check=timestart+window;
timeok=TimeNum()=timestart AND TimeNum()=Check;
Buy= Buyok AND timeok;
Sell= Sellok;
Short= Shortok AND timeok;
Cover= Coverok;
ApplyStop(stopTypeLoss,stopModePoint,sl*ATR(14)); //9
ApplyStop(stopTypeProfit,stopModePoint,tp*ATR(14)); //1.2
Equity(1);
System2:
percentrisked=0.007;
sl=4.5;
tp=2.5;
SetOption(maxopenpositions,1);
CCIperiod=Optimize(CCI,36,34,40,2);
CCIthreshold=Optimize(CCIthres,97,88,96,2);
MAperiod= Optimize(maperiod,7,7,9,1);
TimeFrameSet(inHourly);
MA1=MA(C,MAperiod);
MA2=MA(Ref(C,-0),MAperiod);
CCIhr= CCI(CCIperiod);
ATR1= ATR(14);
TimeFrameRestore();
number=((percentrisked)/(TimeFrameExpand(Ref(atr1,-0),inHourly)*sl));
SetPositionSize(number,spsPercentOfEquity);
CCIshort=TimeFrameExpand(CCIhr,inHourly)ccithreshold;
CCIbuy= TimeFrameExpand(CCIhr,inHourly)-CCIthreshold;
Crossup=Cross(TimeFrameExpand(MA1,inHourly),TimeFrameExpand(Ref(MA2,-2),inHourly));
Crossdown=Cross(TimeFrameExpand(Ref(MA2,-2),inHourly),TimeFrameExpand(MA1,inHourly));
Buyok=Ref(CCIbuy,-5) AND Crossup;
Sellok=CCIshort;
Shortok=Ref(CCIshort,-5) AND Crossdown;
Coverok=CCIbuy;
timestart=2;
window=17;
Check=timestart+window;
timeok=TimeNum()=timestart AND TimeNum()=Check;
Buy=Buyok AND timeok;
Sell= Sellok OR CCIexit;
Short= Shortok AND timeok;
Cover= Coverok OR CCIexit;
ApplyStop(stopTypeLoss,stopModePoint,sl*TimeFrameExpand(Ref(ATR1,-1),inHourly));
ApplyStop(stopTypeProfit,stopModePoint,tp*TimeFrameExpand(Ref(ATR1,-1),inHourly));
Equity(1);
Thanks a lot for your suggestions,
Matthias
RE: [amibroker] Backtest multiple systems across multiple timeframes
Hi Ed,
Good to hear from you again. Basetime interval of the databse is 5min (I
don't use ticks, it s downloaded data from my broker, not IB), but both
systems are 15min timeframe, should be easy to handle.
Win XP, SP3, 4GB RAM..32bit OS,
I could create another database, true, but I wanted to backtest the systems
using One/Single - Equitypool logic, all taking their money from ONE
ACCOUNT as done in my RT Trading. Purpose of this is to find out how much
money every sub-system can trade to result in the smoothest equity curve, so
no extensive search or optimization is necessary as the systems are already
profitable. Having 8 systems, it would require 8-optimization variables,
that's true. But because the systems operate on multiple timeframes, I'm not
sure how to re-write the logic. values/results always differ by more than
10%. or AB crashes when optimizing 'percentrisked'.
From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf
Of Edward Pottasch
Sent: Mittwoch, 8. September 2010 22:03
To: amibroker@yahoogroups.com
Subject: Re: [amibroker] Backtest multiple systems across multiple
timeframes
Matthias,
will have a look at your questions in more detail tomorrow. But that
crashing may be caused due to memory problems of your computer rather than
an error in your code.
So if your systems are to be used in the 5-min timeframe try using a
database that has its base time interval set to 5 minutes rather than ticks
or 1-minute. So, if you have a data provider then simply create a new
database using for the base time interval 5 minutes
that solved a lot of my crash problems,
regards, Ed
From: Matthias mailto:meridian...@gmx.de
Sent: Wednesday, September 08, 2010 8:55 PM
To: amibroker@yahoogroups.com
Subject: [amibroker] Backtest multiple systems across multiple timeframes
Hi,
thanks to the contribution of Ed Pottasch, supported by Bruce, I was able to
dig a little deeper into Amibroker coding. Everybody who is interested in
applying multiple systems on the same underlying simultaneously should look
here, great piece of work:
http://finance.groups.yahoo.com/group/AmiBroker-at/message/5349
Thanks Ed, thanks Bruce.
Unfortunately, I stumbled across a couple of questions when backtesting
multiple systems across different timeframes, hope someone can help, sorry
for the post being a bit lenghty.
Both systems are traded on the same underlying, in order to make things
easier for AB (Which is a bit strange) I used the same set of data, just
renamed it. both systems operate on the same timeframe, say 15mins.
Question 1:
I use the same variable percentrisked for both systems. Wanted to optimize
for percent risked (only!, this is NOT shown in the example below), so to
say capital allocated to each system for the smoothest equity curve, AB
keeps crashing... Can I use the same variable name in each sub-section or
are there limits? should I dedicated percentrisked1 to system1 and
percentrisked2 to system2 only? I am not a programmer, but for my
understanding, both variables are local, so AB should not be crashing...?
Is using Setoption in this context appropriate or would it result in wrong
values?
if(Name()==DAX_CFD_day1)
{
percentrisked=2.0;
factor=Optimize(ATR-Factor,8.5,3,12,0.5);
number=(percentrisked)/(ATR(14)*factor)*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
.systemlogic here
}
if(Name()==DAX_CFD_day)
{
percentrisked=Optimize(Bolli,0.6,0.5,1,0.1);
sl=2;//Optimize(sl,2,2,2.5,0.5);//good:6
number=(percentrisked/(Ref(ATR(14),-1)*sl))*20;
SetPositionSize(number, spsPercentOfEquity);
SetOption(commissionmode,3);
SetOption(Commissionamount,1.2);
SetOption(AllowSameBarExit,True);
SetOption(ActivateStopsImmediately,True);
SetOption(FuturesMode,True);
SetTradeDelays(1,1,1,1);
Equity(1);
... systemlogic here
}
Question 2:
Both systems above use 15min timeframe. Another system is using 1hr
timeframe and is trading FX. I was not able to re-write the logic so that I
could backtest the 3 systems with AA settings 15min timeframe. Any ideas? I
do have about 8 systems, lowest timeframe is 5min, highest timeframe 4hrs.
That would require a lot of re-writing... Am I alone with my I have too
many-systems Problem or am I missing somehting?
original logic in 1hr timeframe:
percentrisked=0.007;
sl=4.5;
tp=2.5;
number=((percentrisked)/(Ref(ATR(14),-0)*sl));
SetPositionSize(number,spsPercentOfEquity);
SetOption(maxopenpositions,1);
CCIperiod=Optimize(CCI,36,34,40,1);
CCIthreshold=optimize(CCIthres,89,88,96,1);
MAperiod=Optimize(maperiod,7,6,8,1);
MA1= MA(C,MAperiod);
MA2= MA(Ref(C,-2),MAperiod);
CCIshort=CCI(CCIperiod)=ccithreshold;
CCIbuy= CCI(CCIperiod)=-CCIthreshold;
Buyok=Ref(CCIbuy,-1) AND Cross(MA1,MA2);
Sellok=CCIshort;
Shortok=Ref(CCIshort,-1) AND Cross(MA2,MA1);
Coverok=CCIbuy
RE: [amibroker] Re: Optimizer target with genetic optimization
Hi, Mike is right, just change the target in the settings/walk-forward-tab. Anyways, you can design your custom metric, too, this is a very valuable feature, just google amibroker add custom metric. For me, I played along with the max. k-ratio and min. ulcer-index. My daytrading systems on futures and FX became much more stable in out-of-sample tests when optimizing for k-ratio or Ulcer-Index instead of CAR/MDD, especially when keeping the in-sample-period very small, say 3-6months, the out of sample to about 3-6months, too. If you're playing with this feature, I d appreciate it if you could write a little bit about your experience, like what kind of system you trade, what underlying and of course, your optimization target. Thanks, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of radmobile_radmobile Sent: Montag, 23. August 2010 21:48 To: amibroker@yahoogroups.com Subject: [amibroker] Re: Optimizer target with genetic optimization This can be achieved through the custom backtester interface. See: http://www.amibroker.com/kb/2008/05/19/historical-portfolio-backtest-metrics / -RM --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com , DougOriard d...@... wrote: When running one of the genetic optimizers, like cmae or trib, is it possible to change the optimization target from CAR/MDD to just % profit, or other target value? CAR/MDD is too heavily biased on minimizing MDD. It would be better to optimize for CAR*(100 - MDD), where MDD is taken as absolute value. Is there a way to do this?
RE: [amibroker] Re: How to configure CAME to look for a Optimization target MINIMUM
Hi,
Worked for me! Thanks a lot.
Taking my example, I typed Ulcer-Index-II and it did the run. I didn't
understand why it would not show up in the drop-down menu.
Thanks,
Matthias
From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf
Of Mike
Sent: Montag, 9. August 2010 08:16
To: amibroker@yahoogroups.com
Subject: [amibroker] Re: How to configure CAME to look for a Optimization
target MINIMUM
The drop list from which to choose the metric for optimization is an
editable field. Just type in the name (case sensitive) of your metric
instead of choosing from the default list of metrics.
Mike
--- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com ,
Matthias meridian...@... wrote:
Hi,
I tried this one out, but when performing Walk-Forward Analysis, I cannot
choose my custom metric as objective function. Effectively, I'd like to
MINIMIZE Ulcer-Index when performing WFA. Portfolio: enable CBT is ticked,
path is specified. Metric appears in the report, and the column is added,
but cannot be chosen from Walk-Forward Optimization Target.
SetCustomBacktestProc();
if( Status(action) == actionPortfolio )
{
bo = GetBacktesterObject();
bo.Backtest(); // run default backtest procedure
st = bo.GetPerformanceStats(0); // get stats for all trades
Ulcer = -1*st.GetValue(UlcerIndex);
bo.AddCustomMetric( Ulcer-Index-II,Ulcer );
}
Also checked the links where it states:
The Optimization target field defines the optimization raport COLUMN
NAME that
will be used for sorting results and finding the BEST one. Any built-in
column can be used
(as appears in the optimization output), or you can use any custom metric
that you define
in custom backtester. The default is CAR/MDD, you can however select any
other built-in metric from the combo.
You can also TYPE-IN any custom metric that you have added via custom
backtester interface.
Any suggestions?
Matthias
--- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com ,
Mike sfclimbers@ wrote:
Add your own custom metric to be used as the optimization target. Simply
calculate your metric as the negative of your intended metric. The result
will be that originally large values become smaller (i.e. more negative) and
originally small values become larger (i.e. less negative).
Refer to custom metrics for more detail:
http://www.amibroker.com/guide/a_custommetrics.html
Set custom target on WFA settings (applies to regular optimization too,
not just walk forward):
http://www.amibroker.com/kb/2008/02/12/getting-started-with-automatic-walk-f
orward-optimization/
Mike
--- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com ,
elizabeth19852002 elizabeth.kennedy@ wrote:
Hello all,
I would appreciate if someone can please tell me how to configure
optimizations like CAME so that it looks for a MINIMUM (as opposed to a
Maximum) of an optimization target (such as user defined objective
function).
Just by adding OptimizerSetEngine(cmae); in AFL and setting the
Optimization target in the Walkforward tab does not tell it to look for a
MINIMUM or MAXIMUM of the optimization target during optimization ???
Thanking you
Liz
[amibroker] Re: How to configure CAME to look for a Optimization target MINIMUM
Hi,
I tried this one out, but when performing Walk-Forward Analysis, I cannot
choose my custom metric as objective function. Effectively, I'd like to
MINIMIZE Ulcer-Index when performing WFA. Portfolio: enable CBT is ticked,
path is specified. Metric appears in the report, and the column is added, but
cannot be chosen from Walk-Forward Optimization Target.
SetCustomBacktestProc();
if( Status(action) == actionPortfolio )
{
bo = GetBacktesterObject();
bo.Backtest(); // run default backtest procedure
st = bo.GetPerformanceStats(0); // get stats for all trades
Ulcer = -1*st.GetValue(UlcerIndex);
bo.AddCustomMetric( Ulcer-Index-II,Ulcer );
}
Also checked the links where it states:
The Optimization target field defines the optimization raport COLUMN NAME
that
will be used for sorting results and finding the BEST one. Any built-in column
can be used
(as appears in the optimization output), or you can use any custom metric that
you define
in custom backtester. The default is CAR/MDD, you can however select any other
built-in metric from the combo.
You can also TYPE-IN any custom metric that you have added via custom
backtester interface.
Any suggestions?
Matthias
--- In amibroker@yahoogroups.com, Mike sfclimb...@... wrote:
Add your own custom metric to be used as the optimization target. Simply
calculate your metric as the negative of your intended metric. The result
will be that originally large values become smaller (i.e. more negative) and
originally small values become larger (i.e. less negative).
Refer to custom metrics for more detail:
http://www.amibroker.com/guide/a_custommetrics.html
Set custom target on WFA settings (applies to regular optimization too, not
just walk forward):
http://www.amibroker.com/kb/2008/02/12/getting-started-with-automatic-walk-forward-optimization/
Mike
--- In amibroker@yahoogroups.com, elizabeth19852002 elizabeth.kennedy@
wrote:
Hello all,
I would appreciate if someone can please tell me how to configure
optimizations like CAME so that it looks for a MINIMUM (as opposed to a
Maximum) of an optimization target (such as user defined objective
function).
Just by adding OptimizerSetEngine(cmae); in AFL and setting the
Optimization target in the Walkforward tab does not tell it to look for a
MINIMUM or MAXIMUM of the optimization target during optimization ???
Thanking you
Liz
RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis
Hi Lionel, With “this book”, I suppose you mean this one: http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8s=booksqid=1280174971sr=8-1 s=booksqid=1280174971sr=8-1 I was talking to Mike and he said that Tharp uses MC for position sizing in his book “the definitive guide to position sizing”. That’s why Mr. Tharp “entered the discussion”. I just wanted to add a critical review on his new book “supertrader”. So, I guess I don’t understand your question. The books cannot really be compared. Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Lionel Issen Sent: Donnerstag, 29. Juli 2010 03:29 To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Thanks. How does this book compare with Van Tharp’s book? http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8s=booksqid=1280174971sr=8-1 s=booksqid=1280174971sr=8-1
RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis
MC = Monte Carlo Cheers, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Lionel Issen Sent: Freitag, 30. Juli 2010 02:02 To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Hi Matthias: Thanks for this information. Excuse my ignorance, but what does MC mean? Lionel From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Matthias K. Sent: Thursday, July 29, 2010 10:49 AM To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Hi Lionel, With “this book”, I suppose you mean this one: http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8s=booksqid=1280174971sr=8-1 s=booksqid=1280174971sr=8-1 I was talking to Mike and he said that Tharp uses MC for position sizing in his book “the definitive guide to position sizing”. That’s why Mr. Tharp “entered the discussion”. I just wanted to add a critical review on his new book “supertrader”. So, I guess I don’t understand your question. The books cannot really be compared. Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Lionel Issen Sent: Donnerstag, 29. Juli 2010 03:29 To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Thanks. How does this book compare with Van Tharp’s book? http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8s=booksqid=1280174971sr=8-1 s=booksqid=1280174971sr=8-1
RE: [amibroker] Multiple Time Frames
Hi, There's lots of mistakes in this code: 1. You're missing timeframeexpand 2. IIF cannot be use like this, I wonder if it returned you a result at all Ok. I'm not sure about your Stochastic-Thingie, but you should be able to customize my example to your needs. Furthermore check the tutorial. http://www.amibroker.com/guide/h_timeframe.html 5min-Chart, your smallest interval, could be anything, 15min, 1hr, I take 5min for now. I wanna buy if the Close (5min) is greater than EMA(C,200) indaily and only before 12 o'clock in the morning. TimeFrameSet(inDaily); EMA1=EMA(C,200); TimeFrameRestore(); Buy=IIf(CTimeFrameExpand(EMA1,inDaily) AND TimeNum()=12,1,0); Sell=0; Anyways, you don't really need IIF here. Works like this, too TimeFrameSet(inDaily); EMA1=EMA(C,200); TimeFrameRestore(); Buy=CTimeFrameExpand(EMA1,inDaily) AND TimeNum()=12; Sell=0; Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Jonathan Lee Sent: Dienstag, 27. Juli 2010 19:47 To: amibroker@yahoogroups.com Subject: [amibroker] Multiple Time Frames Hi amibroker experts, I'm a little stuck with multiple time frames - and was hoping someone could assist. It's probably a straight forward solution as I'm pretty much a newbie at this. I've got a basic oscillator (similar to stochastics) in which I have determined that 30 is oversold, 70 is overbought and 50 at mid range. i'm using the higher time frame as a condition, so looking at the hourly chart, I only want a buy signal to be actioned if the daily chart (%K and %D) is less than the mid range (50). I put all my daily parameters between: TimeFrameSet (inDaily) and TimeFrameRestore (); And my buy/sell instruction (in the hourly setting) is as follows: IIf((DailySK=DailyMid AND DailySD=DailyMid),Buy=SK=Lower AND SD=Lower AND Cross(SK,SD),0); IIf((DailySK=DailyMid AND DailySD=DailyMid),Sell=SK=Upper AND SD=Upper AND Cross(SD,SK),0); Short=Cover=0; For some reason I am getting results which ignore the higher time frame? Much appreciated. Jonathan
RE: [amibroker] Optimize Time Parameters
Hi, I'm using this. It's pretty much self-explanatory. Works based on hours. Depending on which data you use, the timestamp might be different. So to say timestart might be 055900 instead of 06. Watch out for Fridays, too, if you don't wanna hold the trader over the weekend. timestart=06;//Optimize(timestart,06,00,23,1); window=08;//Optimize(windowlength,07,04,10,1); Check=timestart+window; timeok=TimeNum()=timestart AND TimeNum()=Check; Greetings, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of pcmoxon Sent: Mittwoch, 28. Juli 2010 17:11 To: amibroker@yahoogroups.com Subject: [amibroker] Optimize Time Parameters Hi, I am working on an intraday system and want to optimize the trading hours the system should trade. I have a Start Time parameter (using ParamTime) and I would like to optimize the system using this field. Does anyone know how to optimize using Time fields? Any examples you know of would really help. Kind Regards, Pete
RE: [amibroker] Re: Trading Systems, Position Sizing and Monte Carlo Analysis
Hi, Thanks Mike for your explanation and the link. The definite guide to position sizing is the only book I'm missing in my van Tharp collection, looks like just found my next bday present, hehee. Just to make this complete, the new Supertrader book is pretty lame in my opinion, nothing really new, just a little bit of everything from Tharp. I shall look up on Vince, thanks for the info. Following up with the previous posts we need to look outside the samples to get the full pictures I'd like to hear a little more on that, too, if you're willing to share. Thanks, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of mkecera Sent: Mittwoch, 28. Juli 2010 17:48 To: amibroker@yahoogroups.com Subject: [amibroker] Re: Trading Systems, Position Sizing and Monte Carlo Analysis Yes. That's what I was trying to point out. I would be interested to hear more about we need to look outside the samples to get the full pictures if you would like to share. Best regards, MK --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com , paultsho paul.t...@... wrote: I think the emphasis of MK's comments is about the use (or misuse) of the phase monte carlo simulation to equate to random rearrangement of trades and its possible ramification. Monte Carlo simulation is based on ARTIFICALLY simulating data points by recreating the a chance process, running it many times, and directly observing the results - http://www3.wabash.edu/econometrics/EconometricsBook/chap9.htm bootstrap is resampling repeatedly and randomly from an original, initial sample - http://www3.wabash.edu/econometrics/EconometricsBook/chap23.htm. While leaving the discussion on assumptions aside (I'll say more when I get a bit more time). One of the problems of equating bootstrapping with monte carlo is that incorrect conclusions are also drawn about the methodology. I do not believe resampling of trades gives us a reliable picture about the behaviour of a trading system. We should all be familiar with rearranging the deck chairs on the Titanic does no good at all. Similarly, I think we need to look outside the samples to get the full pictures.
RE: [amibroker] Re: Trading Systems, Position Sizing and Monte Carlo Analysis
Hi, Mike, you suggested Ralph Vince - The Handbook of Portfolio Mathematics: Formulas for Optimal Allocation Leverage (Wiley Trading 2007) - The Leverage Space Trading Model: Reconciling Portfolio Management Strategies and Economic Theory (Wiley Trading 2009) Did you read them? Could you give me a brief summary, something a little bit more critical than Amazon’s praises to the sky? I like Vince’s work, guess I shall read them, I just don’t like the term economic theory. But before this topic falls asleep: If I got it right: This one is about trading one system on a market-portfolio, than figuring out some statistics etc. What I’d like to read up on is something like a TRADING SYSTEM PORTFOLIO on 1 market, so to say not pick the markets scientifically, but the systems. Additionally, my comments on MC are incomplete because I didn’t dig very deep into the material. I am aware of the methods you described below, but still I don’t really believe it’s useful. What’s the use of knowing average performance, best performance, worst performance, etc.? If you have that info, do you base something on it, say position sizing, emergency system stop? My walkforward oos is something that I do trust more, guess that’s my opinion. From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Mike Sent: Dienstag, 27. Juli 2010 00:02 To: amibroker@yahoogroups.com Subject: [amibroker] Re: Trading Systems, Position Sizing and Monte Carlo Analysis Books that you might be interested in: Ralph Vince - The Handbook of Portfolio Mathematics: Formulas for Optimal Allocation Leverage (Wiley Trading 2007) - The Leverage Space Trading Model: Reconciling Portfolio Management Strategies and Economic Theory (Wiley Trading 2009) Your comments on Monte Carlo are incomplete. Monte Carlo is not just changing the the order of trades. Monte Carlo can also be used to treat existing trades as a pool of trades from which a greater number of trades may be sampled in order to project a longer term equity curve. For example, if you have a collection of trades that you feel adequately represent the range of trades that your system will generate, you can randomly draw some number of trades (with replacement) from the pool to generate thousands of equity curves (e.g. draw 100 trades from a pool of 30). You may then analyze the resulting equity curves to get such information as average performance, best performance, worst performance, etc. Mike --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com , Matthias K. meridian...@... wrote: Hi, Indeed this is a very interesting topic. Many thanks go to Howard Bandy, it is literally a must read when working with Amibroker. I’m curiously waiting for the new one. Regarding System Design and Position Sizing: Well, I believe that there are a lot of books out there, but different authors really do mix up terminology, so this is how I interpret it: Position sizing Tells you how much, e.g. how many contracts or shares one should buy, Howard just mentioned 2 algos, fixed fraction and fixed ration. Ralph Vince’s book is a must read on this, however, it’s not quite easy going. Furthermore, there’s martingale and antimartingale, whereas martingale pretty much means: the more you lose, the more you bet. I believe it’s a sure-fire way to the poorhouse, that’s why I use ALWAYS anti-martingale. No black swans welcome in my trading. Risk management Obivously, there’s a clear overlap with position sizing, but still I believe it’s meant to be a little different. Say you have a system with a plugged in position sizing algo. You backtest it and you’ll have an estimate of the drawdown. Obviously, you will want to adjust position size so that you can stomach the drawdown. But as a safety measure, you might plug in another strategy. It might be an equity MA-Crossover, it might be something psychological (a divorce, an illness), it might also be a safety stop due to increased market volatility / markets crash. Remember to always measure drawdown in percentage terms, e.g. Tradestation Strategy Report is pretty useless, because it only shows drawdown as an absolute figure so it requires Excel to actually calculate drawdown in percentage terms. Also CAR is calculated correctly in AB, I’ve seen other software, costing 2k€, still calculating CAR wrong (10 times 10% per annum is NOT 100%) Portfolio Management Hahaa, I believe this is where a few gold bucks a buried. Portfolio in terms of which markets to trade has been discussed in the literature, still this is a very specialized topic already. There not many books found on this, that’s why I spend so much time thinking about it: Nothing to be found means a lot to be made and understanding it in great detail might give me/you a larger edge. Especially
RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis
estimate of a historic drawdown, but I achieve the same by multiplying my non-monte-carlo-drawdown with say 1.5… Greetings from Germany, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Howard B Sent: Montag, 26. Juli 2010 16:33 To: amibroker@yahoogroups.com Subject: Re: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Hi Sohamdas -- In my opinion, this is definitely a topic that deserves discussion in the AmiBroker forum. What position sizing should be used during backtests? If you will be evaluating each trade for its characteristics -- entry efficiency, exit efficiency, and so forth, then each trade should be the same size. For stocks and ETFs that means the same dollar amount. For futures that means the same number of contracts. If you will be comparing equity growth over a period of time to other alternatives, then you will want position sizing and / or compounding to some degree. For example, if you want to compare the results of a trading system to buy and hold, you will want to take the same size position at the beginning if the test period for each alternative, then compare equity smoothness, growth, drawdown, etc. If you are planning to use aggressive position sizing, there are several things to consider. 1. I cannot state it too often -- your system must have a positive expectancy measured on strictly out-of-sample results. You absolutely cannot use in-sample results to estimate the likely future performance of a trading system in any event. And if aggressive position sizing is based on in-sample results, you will go bankrupt. 2. Traders should have a business plan in place. They should know when to quit -- either when they have enough that they no longer need to trade, or when they have lost so much that they can no longer trade or realize that they should pick another profession. 3. Aggressive position sizing depends on having: A. Positive expectancy. B. Understanding of risk. Both the risk that is acceptable for each trade from the account, and the risk associated with the trading system. Most trading systems have higher risk per trade than the account risk allows, so even taking a position that is everything you can afford to buy is aggressive. C. The ability to use leverage. Brokers allow use of margin, and some ETFs have leverage. By using these, it is possible to get 12 to 20 times leverage trading stocks and ETFs in an ordinary brokerage account. D. Frequent trading, because that provides frequent compounding. For most trading systems, the final equity of an account is a multiple of the initial equity that can be computed from: terminal_equity = (1 + expectancy) ^ number_of_trades where expectancy is the average percentage gain per trade. There are two general schemes for aggressive position sizing. As you dig into the math, you will see that they are closely related. The first is fixed fraction, popularized by Ralph Vince. The second is fixed ratio, popularized by Ryan Jones. Both men have written books and papers describing their methods, and you can do an Internet search on each phrase and get a lot of information. The essence of both methods is to increase position size when the system is operating profitably. In gambling terminology, you are betting the run of the table. When winning, increase; when losing, pull back. Both of these are betting schemes called anti-martingale. Ralph Vince has also popularized the notion of optimal f -- that fixed fraction that should be bet on each play to maximize the terminal equity. The fraction of the account used for each play is determined by the largest anticipated drawdown or trade loss. He, and everyone else who is working with real money, shows that the fraction bet on each play Must be less than optimal f if the account is to remain solvent. In fact, the fraction must be much less than optimal f if the account is avoid large drawdowns. Trading at optimal f essentially guarantees drawdowns in the 80 percent range. Ryan Jones essentially creates two sub-accounts. One is the original stake, say $100,000. The other is the profits from trading. Ryan waits until there are some profits, then uses a high percentage of the profits for each trade. The two methods converge mathematically. Some traders prefer to begin using one method, then switch to the other as profits accumulate. Before you consider using an aggressive position sizing scheme, and buying all the stock you can afford is aggressive, please read my other comments in both this forum and Aussie Stock Forums related to trading system development and position sizing. And read both Vince's and Ryan's books, and other material you can find on the Internet. The input to the simulators that model either fixed fraction or fixed ratio need is a list of closed trade results. These are individual trades, each
RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis
Well, it’s written in English, otherwise it wouldn’t be very useful to post it ;) Try this one instead. http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-Optimisation-Portfolio-Construction/dp/1905641796/ref=sr_1_1?ie=UTF8s=booksqid=1280174971sr=8-1 s=booksqid=1280174971sr=8-1 If living in the States, I believe you could order it from amazon UK, I’m sure they have it in stock there. Just please be aware that it scratches this topic (multiple systems portfolio, portfolio composition, etc.) only. Half of the book is “beginner’s” talk, like where to start from and so forth. The remaining rest is, very very good food for thought. I don’t quite understand the critics in Amazon. (Ok, ok enough advertisement J ) Maybe this one is worth adding (Stridsman): http://www.amazon.com/Trading-Systems-That-Work-Evaluating/dp/007135980X/ref=sr_1_1?ie=UTF8 http://www.amazon.com/Trading-Systems-That-Work-Evaluating/dp/007135980X/ref=sr_1_1?ie=UTF8s=booksqid=1280175191sr=1-1 s=booksqid=1280175191sr=1-1 This book is outstanding from the very page onwards. But it deals with “long-term” trendfollowing, futures being the only instrument traded (I don’t have the money to actively trade say 20 futures markets long-term. To do so, you need 2.5M€ or about 3m$, that’s what I estimated) The methodologies inside are so much worth the money. From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Lionel Issen Sent: Montag, 26. Juli 2010 22:01 To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Your link for Trading Systems: A New Approach to System Development and Portfolio OptimisationTrading Systems: A New Approach to System Development and Portfolio Optimisation led to Amazon Germany. I couldn’t find any US sellers. Is the book only published in German, is it also available in English? Lionel Issen From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Matthias K. Sent: Monday, July 26, 2010 2:19 PM To: amibroker@yahoogroups.com Subject: RE: [amibroker] Trading Systems, Position Sizing and Monte Carlo Analysis Hi, Indeed this is a very interesting topic. Many thanks go to Howard Bandy, it is literally a must read when working with Amibroker. I’m curiously waiting for the new one. Regarding System Design and Position Sizing: Well, I believe that there are a lot of books out there, but different authors really do mix up terminology, so this is how I interpret it: Position sizing Tells you how much, e.g. how many contracts or shares one should buy, Howard just mentioned 2 algos, fixed fraction and fixed ration. Ralph Vince’s book is a must read on this, however, it’s not quite easy going. Furthermore, there’s martingale and antimartingale, whereas martingale pretty much means: the more you lose, the more you bet. I believe it’s a sure-fire way to the poorhouse, that’s why I use ALWAYS anti-martingale. No black swans welcome in my trading. Risk management Obivously, there’s a clear overlap with position sizing, but still I believe it’s meant to be a little different. Say you have a system with a plugged in position sizing algo. You backtest it and you’ll have an estimate of the drawdown. Obviously, you will want to adjust position size so that you can stomach the drawdown. But as a safety measure, you might plug in another strategy. It might be an equity MA-Crossover, it might be something psychological (a divorce, an illness), it might also be a safety stop due to increased market volatility / markets crash. Remember to always measure drawdown in percentage terms, e.g. Tradestation Strategy Report is pretty useless, because it only shows drawdown as an absolute figure so it requires Excel to actually calculate drawdown in percentage terms. Also CAR is calculated correctly in AB, I’ve seen other software, costing 2k€, still calculating CAR wrong (10 times 10% per annum is NOT 100%) Portfolio Management Hahaa, I believe this is where a few gold bucks a buried. Portfolio in terms of which markets to trade has been discussed in the literature, still this is a very specialized topic already. There not many books found on this, that’s why I spend so much time thinking about it: Nothing to be found means a lot to be made and understanding it in great detail might give me/you a larger edge. Especially a scientific approach that is not based on gut-feel, is missing. This is what I am currently working on: Portfolio Management in terms of Trading Systems. So to say, have a trading systems farm, probably 3-5 systems on one underlying to have a very smooth equity curve. This includes diversification across a) methodology (Mean Reversion, Trendfollowing, Breakouts,etc.) and b) timeframe (day-trading, swing trading
RE: [amibroker] Help with DDE connections using TOS
Hi, Did you try putting a symbol into realtime quote list by type in symbol (use the symbol's name of your DDE provider). Other than that, try to look up in the documentation of TOC, not sure what it stands for/where it comes from. NO backfill available via DDE. Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of bar607 Sent: Donnerstag, 22. Juli 2010 23:11 To: amibroker@yahoogroups.com Subject: [amibroker] Help with DDE connections using TOS Hi, I configured the DDE connection according to this post with people that have successful done it http://finance.groups.yahoo.com/group/amibroker/message/149673 I did exactly as it stated but I still can NOT connect through DDE. Can anyone help me? Thanks, Pin
RE: [amibroker] Re: Help with DDE connections using TOS
Hi, Backfill means filling your chart with historical data for a specific period. All you can do with DDE is to get CURRENT realtime quotes, store them and good. If you switch off your computer on Monday, turn it on on Tuesday, you won't have data for Monday, no way to get it from your DDE source/interface. However, you can import historical quotes MANUALLY if your server has historical data available in ASCII format, e.g., and you can download them. With your provider, this seems to be the case, but it's quite a hassle updating the database manually if trading say, 1min charts. An elegant solution is the COM interface, that several providers, e.g. Esignal offer. AB has a dedicated Plug-In for each provider in the list. Backfilling data with the COM/RT-plugin is not a problem, however, it'll require you to spend a little money on a monthly base. Have a nice weekend, Matthias From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of bharat Sent: Freitag, 23. Juli 2010 19:53 To: amibroker@yahoogroups.com Subject: [amibroker] Re: Help with DDE connections using TOS Dear Matthias You have written that no backfill available via DDE.But does it apply to server which has data available?I think server of TBMX has 14 years data stored on it,because I am able to draw chart for 14 yeas data. --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com , Matthias K. meridian...@... wrote: Hi, Did you try putting a symbol into realtime quote list by type in symbol (use the symbol's name of your DDE provider). Other than that, try to look up in the documentation of TOC, not sure what it stands for/where it comes from. NO backfill available via DDE. Matthias From: amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com [mailto:amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com ] On Behalf Of bar607 Sent: Donnerstag, 22. Juli 2010 23:11 To: amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com Subject: [amibroker] Help with DDE connections using TOS Hi, I configured the DDE connection according to this post with people that have successful done it http://finance.groups.yahoo.com/group/amibroker/message/149673 I did exactly as it stated but I still can NOT connect through DDE. Can anyone help me? Thanks, Pin
[amibroker] Who is trading more than 5 Systems simultaneously with AB?
Hi, I'm looking for an approach to run more than only 1 or 2 systems automated with AB. Is there someone in here who can offer advice(hardware, coding, etc.) or has faced the same problem? I have read up on AT-programming and I did grasp the concept, but after going through this forum and a couple of emails to the support, I cannot really find a workable solution. Maybe I'm understanding something wrong, but combining 2 Systems in one AFL seems pretty tricky already. What happens if I get a long and short at the same bar when trading only one underlying!? solutions that were offered are not useful: Position-Pool which sums up the positions before sending them to IB Not to mention the debugging of a 4-Systems-AT-AFL. So, I'm willing to do whatever it takes - please don't get me wrong - if the idea is worthwhile pursuing, but I simply don't have a clear-cut approach/solution for this. Obviously finding some good systems that complement each other is tricky already, spending hours n hours of coding just to combine and debug more than 2 systems, I'm sorry but it'll give me a serious headache. I mean, how would it be like if I had to combine 10 Systems into 1 AFL? Guess the outcome would kinda like look scary. Is running several instances of AB connected to one IB account a good idea? Thanks in advance, Greetings from Germany, Matthias
[amibroker] Re: Who is trading more than 5 Systems simultaneously with AB?
Sorry, this was meant to be in the Autotrading Forum, mixed it up. Anyways, it's in this one and the at-forum now. Didn't mean to be pushy. M
RE: [amibroker] Backtest Report Long/Short trades columns?
Hi, It's very hard to understand the system's logic only by numbers. But as a matter of fact you identified the problem already: when you're testing long only, the shorts are skipped, thus it'll result in a very different equity curve as compared to the result when trading long and short together. So far, if everything went right, you just figured out the difference of trading a long-only system and a short-only system. If trading stocks, one might consider the long-side only because the indices and stocks sort of have a long-bias. In order to smoothen your equity curve, you might want to combine a short and a long system. The result is something like you're showing here. It'll pretty much enable you to make money in both bull and bear markets and ideally smoothens out the equity curve and reduces drawdowns and flat periods. Looks good to me if your coding has been right and doesn't look into the future. Always: CAR/MDD ratio: above 1,5 over 10 years time is outstanding, as it's a risk adjusted measure, it'll equal/weigh up profits versus losses|drawdown and will not consider net profit only To sum it up: the more trades you have, the longer your backtesting period is, the higher your CAR/MDD ratio: the better your system. Don't forget to include commissions/ spreads, anyhow. They might deteriorate your big picture. Especially with mid-caps. A good custom backtest metric might be a combination of number of trades and CAR/MDD, so to say, make 50 Trades per year and still give a CAR/MDD ratio above 1. Greetings M From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of Ron Sent: Donnerstag, 22. Juli 2010 06:31 To: amibroker@yahoogroups.com Subject: [amibroker] Backtest Report Long/Short trades columns? I have a strategy that over time is either long or short but not both at the same time. I'm having difficulty understanding the Long and Short trades columns in the back test report when I run the back test in Long and Short mode. Long only backtest gives (Net Profit 458.76%) Initial capital 10.00 10.00 10.00 Ending capital 558757.12 558757.12 106586.98 Net Profit 458757.12 458757.12 6586.98 Net Profit % 458.76 % 458.76 % 6.59 % Exposure % 38.28 % 38.28 % 0.00 % Net Risk Adjusted Return % 1198.47 % 1198.47 % N/A Annual Return % 30.78 % 30.78 % 1.00 % Risk Adjusted Return % 80.42 % 80.42 % N/A Short only backtest gives (Net Profit 75.89%) Initial capital 10.00 10.00 10.00 Ending capital 175889.18 106586.98 175889.18 Net Profit 75889.18 6586.98 75889.18 Net Profit % 75.89 % 6.59 % 75.89 % Exposure % 28.44 % 0.00 % 28.44 % Net Risk Adjusted Return % 266.88 % N/A 266.88 % Annual Return % 9.21 % 1.00 % 9.21 % Risk Adjusted Return % 32.38 % N/A 32.38 % The Long Short backtest gives, Initial capital 10.00 10.00 10.00 Ending capital 921758.81 712441.67 315904.12 Net Profit 821758.81 612441.67 215904.12 Net Profit % 821.76 % 612.44 % 215.90 % Exposure % 66.78 % 38.28 % 28.50 % Net Risk Adjusted Return % 1230.56 % 1599.91 % 757.56 % Annual Return % 41.40 % 35.84 % 19.65 % Risk Adjusted Return % 62.00 % 93.62 % 68.96 % Notice the Long Net Profit is 612.44% and Short is 215.90% which is quite a bit different then when I ran them independently. Other columns are different as well but I'll focus on Net Profit % for now. I expect there is some compounding thing happening here but if someone asked me I couldn't explain it. Does anyone out there have a explanation. Thanks in advance, Ron
RE: [amibroker] Still trying to combine systems
Hi,
Any news on your approach? Just ran through the afl online lib and found an
idea from Paul Ho called something like testing multiple systems without
going into CBT. Maybe you wanna give it a try
Greets from Germany
M
From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf
Of Gonzaga
Sent: Mittwoch, 14. Juli 2010 17:49
To: amibroker@yahoogroups.com
Subject: [amibroker] Still trying to combine systems
Hi
I have been fighting with AFL to write a sample program that let us combine
two systems, as I told in a recent thread.
I haven´t been able.
The idea is easy: two type of buy conditions, one in the crossing of two
averages, the other after falling four bars, using EOD bars.
Both systems uses the same pool of money, and are PORTFOLIO SYSTEMS, buying
as far as ten stocks, 10% of the money every purchase.
THE BUY condition is easy to combine. But I have to tell AFL that if the
last BUY as been using the First condition, the SELL has to be checked with
the crossing of the averages again.
And if the last BUY condition as been the second, the SELL condition has to
be another one. (2 bars away from buying day).
After trying it many times without using Custom backtester interface, I
think the only way is using backtester interface.
So I write the exact code, so anybody can test it and fill the custom
backtest part..
I think it should be easy; the problem is that I don't really understand the
logic of the custom backtester, so I don't know what to do
Any suggestion will be very very much appreciated..
//BEGGINING OF CODE
posQty=Param(Posiciones simultaneas,10,1,20,1);
CapMinimo=Param(Cap minimo,2,1,200,5);
SetOption(InitialEquity, 10 );
SetOption(AllowPositionShrinking, True );
SetOption(MaxOpenPositions, PosQty );
SetOption (accountmargin,100);
PositionSize = -100/(posqty);
Buy=Sell=0;
SetTradeDelays(1,1,1,1);
//FIRST SYSTEM---
Buy1= Cross(MA(C,15),MA(C,70));
BuyPrice1=Open;
SellA=Cross(MA(C,70),MA(C,15));
SellPriceA=Open;
//SECOND SYSTEM---
Buy2= CO AND Ref(C,-1)Ref(O,-1) AND Ref(C,-2)Ref(O,-2) AND
Ref(C,-3)Ref(O,-3);
SellB= Ref(Buy,-2);
//BUY-
Buy=IIf(Buy1,True,IIf(Buy2,True,False));
BuyPrice=Open;
nombre1=Name();
typeBuy=IIf(Buy1,1,IIf(Buy2,2,0));//if buy1, array 'typebuy' is 1. If Buy2,
is 2. Else is 0
n=BarsSince(Buy);
//NEXT TWO SENTENCES DOES NOT WORK
//Sell= IIf( Ref(typebuy,-n)==1, Cross(MA(C,15),MA(C,70)),False);
//Sell= IIf( NOT Sell AND Ref(Buy2,-n), Ref(Buy,-2),False);
//INSTEAD, USE OF CUSTOMBACKTESTER
SetCustomBacktestProc();
if (Status(action) == actionPortfolio) {
bo = GetBacktesterObject(); // Get backtester object
bo.PreProcess(); // Do pre-processing (always required)
for (i = 0; i BarCount; i++) // Loop through all bars
{
for (sig = bo.GetFirstSignal(i); sig; sig = bo.GetNextSignal(i))
{
if (sig.isentry() AND Buy1[i])
//m
} // End of for loop over signals at this bar
bo.ProcessTradeSignals(i); // Process trades at bar (always required)
} // End of for loop over bars
bo.PostProcess(); // Do post-processing (always required)
}
PositionScore=Ref(StochK(15,3),-1);
Buy=ExRem(Buy,Sell); Sell=ExRem(Sell,Buy);
PlotShapes(IIf(Ref(Buy,-1),shapeUpArrow,shapeNone),colorBlue,0,L,-15);
PlotShapes(IIf(Ref(Buy,-1),shapeHollowSquare
,shapeNone),colorBlue,0,BuyPrice,0);
PlotShapes(IIf(Ref(Sell,-1),shapeDownArrow,shapeNone),colorRed,0,H,-15);
PlotShapes(IIf(Ref(Sell,-1),shapeHollowSquare,shapeNone),colorRed,0,SellPric
e,0);
//---END OF CODE--
RE: [amibroker] Re: pl. code this
Why not spend a minute in YOUR education and try to solve your programming riddle yourself? If theres specific advice on a specific step in your coding, go ask for it: Please do! Present your code! Theres plenty of sophisticated long-term users in here that offer their valuable advice for free if they see your piece of the pie But as long as you dont want to dig a little deeper into Amibroker coding, why bother at all? Im sure you bought that piece of software because itll serve you a precious service for the time youre trading / future purpose. So why not make the effort and understand it in great detail? Believe me, this path is worth pursuing. Btw: this awesome John Carter Idea is not going to be as great as you think - trading it mechanically. Greets from Germany, M From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of bharat Sent: Montag, 19. Juli 2010 04:06 To: amibroker@yahoogroups.com Subject: [amibroker] Re: pl. code this AFL which you have mentioned is available here.Check and tweak. --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com , cas soni soni...@... wrote: Hello ,I had given you BBsqueeze .afl  indicator long time ago...use that..i think it is same as per mt4 indicator..* i dont have that afl have a nice time. . --- On Sun, 18/7/10, bharat bharat_parth2...@... wrote: From: bharat bharat_parth2...@... Subject: [amibroker] Re: pl. code this To: amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com Date: Sunday, 18 July, 2010, 5:41 PM  Hi Found full MT4 code for this.Pl. Code AFL === // This indicator is based on a strategy mentioned in John Carter's book, Mastering the Trade. The basic idea // behind the strategy is that markets tend to move from periods of low volatility to high volatility and // visa versa. The strategy aims to capture moves from low to high volatility. For gauging this he uses two // common indicators - Bollinger Bands and Keltner Channels (ok, not so common!). He also uses the Momentum // indicator to provide a trade bias as soon as the Bollinger Bands come back outside the Keltner Channels. // // The Squeeze_Break indicator combines this into a signal indicator and has the following components: // 1. A positive green histogram means that the Bollinger Bands are outside the Keltner Channels // and the market is lightly to be trending or volatile. The stronger the histogram the stronger // the directional price move. // 2. A negative red histogram means that the Bollinger Bands are inside the Keltner Channels // and the market is lightly to be consolidating. The stronger the red histogram the tighter // price action is becoming. // 3. Incorporated into the indicator is a Momentum indicator. According to the strategy J. Carter // goes long when the Bollinger Bands break outside the Keltner Bands and the Momentum indicator // is above the zero line. He goes short when the Momentum indicator is below the zero line on the // break. // 4. I've also added other indicator info in the top left hand corner to give a broader idea // of current market conditions. // 5. The indicator provides audio alerts when a potential breakout is occurring. // // This indicator tends to be better with the larger timeframes. Personally I don't trade on an alert // signal alone. It's just a handy tool for warning me of potential breakout trades. // === #property copyright DesORegan #property link mailto: oregan_...@... // // indicator properties // #property indicator_separate_window #property indicator_buffers 3 #property indicator_color1 ForestGreen #property indicator_color2 Red #property indicator_color3 Blue // === // User Inputs // === extern int Boll_Period=20; extern double Boll_Dev=2.0; extern int Keltner_Period=20; extern double Keltner_Mul=1.5; extern int Momentum_Period=12; extern int Back_Bars=1000; extern bool Alert_On=true; extern bool On_Screen_Info=true; // = // Buffer Array Declarations // = double Pos_Diff[]; // Pos Histogram double Neg_Diff[]; // Neg Histogram double Momentum[]; // Momentum Indicator // === // Internal Array Declarations // === double Squeeze[]; // Used to track which i (index value) is above // and below zero line. 0 followed by 1 triggers alert // = //
RE: [amibroker] Hold a trade for a minimum # of bars
Hi,
This one is from Mr. Janeczko, he put it online as an example, but I cannot
find the link. Replace buy/sell,etc. witch your rules, the rest is pretty
much self-explanatory.
Buy = MACD() 0 AND RSI() 30;
Sell = MACD() 0 OR Cross( 70, RSI() );
SetPositionSize(1,spsShares);
// now we would like to ensure that position is NOT
// exited during first MinHoldBars
MinHoldBars = 17; // say don't want to exit for first 17 bars since entry
// first filter out buy signals occuring outside testing range
Buy = Buy AND Status(barinrange);
BarsInTrade = 0;
for( i = 0; i BarCount; i++ )
{
// if in-trade, then increase bar counter
if( BarsInTrade 0 ) BarsInTrade ++;
else
if( Buy[ i ] ) BarsInTrade = 1; // on buy signal start counting
bars-in-trade
// if we are in trade - remove sells occurring too soon
if( BarsInTrade MinHoldBars ) Sell[ i ] = 0;
else
if( Sell[ i ] ) BarsInTrade = 0; // on sell reset barsintrade flag
}
printf(barsintrade:\n);
WriteVal(barsintrade);
Greets,
Matthias
From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf
Of JEFF F
Sent: Donnerstag, 15. Juli 2010 00:41
To: amibroker@yahoogroups.com
Subject: [amibroker] Hold a trade for a minimum # of bars
Can anyone tell me how to hold a trade for a minimum # of bars (such as 3)
before checking for a sell signal?
I have tried settradedelay(0,3,0,0) but when i backtest, the trade still
happens the next day. Same with changing the delay within the settings
button.
Any help is appreciated.
Thanks
RE: [amibroker] FOREX brokers and data feed differences
Not sure if it helps, but give it a try: FX is traded 24 hrs, thus the timestamp of your MT broker may be stamped differently as compared to finam (=Russia) data. Be careful with data selection in FX anyways, it's an interbank market that is not watched by authorities, hence is not regulated. Data from one provider may be more volatile, compared to another. Having too little information about your project, it's difficult to give you better advice than this: When backtesting a FX-System, be sure that the data comes from the broker you want to trade with! If you're thinking of buying data, you might stumble across their site: http://disktrading.is99.com/disktrading/ Unfortunately the FX data is very very noisy and I abandoned it for system development in FX (unless you trade on daily charts). Their futures data is ok (as in ok, not outstanding), though. Here's another free data source you may wanna play with. http://www.dukascopy.com/swiss/english/data_feed/csv_data_export/ Last but not least: IMHO always give ECN brokers a preference, they're interested in executing your trades only and don't act as a market maker. IB is an ECN borker, MB Trading (has MT4) is another one. From: amibroker@yahoogroups.com [mailto:amibro...@yahoogroups.com] On Behalf Of BruceD Sent: Dienstag, 6. Juli 2010 06:59 To: amibroker@yahoogroups.com Subject: [amibroker] FOREX brokers and data feed differences This is my first post on this forum so please point me in the right direction if I'm doing something wrong. I'm looking for advice from someone who may have been down this road allready. I've been trading FOREX using MQ4 for a few years, but have allways been frustrated by its inadequate back testing capabilites. I recently found AmiBroker and Howard Bandy's books. This has been a real quantum leap for me and I'm anxious to apply some of the methods I've learned. Here's the problem: Using AmiQuote to download FOREX history, the data feed comes from FinAm. Using the backtesting and optimization tools in AmiBroker I've develped a succesfull trading system, however, I don't have a broker set up to accept live trades from AmiBroker yet. As a test, I wrote the same system in the MQ4 system with my existing broker and ran it live for a while. Its performance was poor. Further research has shown that difference in performance is due to the fact that the data from FinAm and that from my MQ4 based broker (IBFX) is very different! I would like to switch to a FOREX broker that will allow me to do live trading from AmiBroker, hopefully that uses the same data source as FinAm, and of course that has reasonable rates. If I read the info on FinAm's website properly, their small USD 10 fee per trade in addition to a higher than normal spread is a deal breaker for me. Has anyone else experienced this issue with FOREX data? Does anyone have a broker they are doing live FOREX trading with via AmiBroker that they are happy with? I do a lot of intra-day trades. Any suggestions would be appreciated. Again, if I'm posting incorrectly or in the wrong area, please let me know. Thanks!
[amibroker] Re: How to mix systems
As you can see from my questions/answers, I'm always interested in finding new solutions. but quote me if I'm wrong: with your approach you are NOT using ONE equity pool. Say you have 100k$, with your methodology you would do one of the following: Assign 50k$ or 50% to System1 Assign 50k$ or 50% to System2 == Backtest and combine equity curves as you said. Alternativley you might prop up the result in using 100k$ to backtest your 2 Systems. If you operate on a non-percent-risk model, it might work, but a non-percent-risk model ( e.g. fixed number of shares) is not really state-of-art. ...just thinking out loud... --- In amibroker@yahoogroups.com, notanaiqgenius notanaiqgen...@... wrote: Hi System Mixers, I've been following this thread. I see people want to do different things when they say mix systems. If all you want to do is backtest on daily bars with multiple systems, yet be able to test each system by itself first, one simple way might be to obtain each system's equity curve (preferably walk-forward OOS curve), and then export that curve to a CSV file. Once you have say, 2 or more systems that you want to time / combine, you can then create a new AB database with nothing in it, and then import the equity curves as pseudo/fake symbols using just Date and Close, where the Close is your equity value from that system. Then you could write one code to rank or time the equity curves. Sure, that's quite a bit different than controlling and combining all the individual orders of all the systems, but if you want to keep your system's separate and just time their individual equity curves, then this sort of approach might work. The down side is that you wouldn't be able to see how much risk you are taking on 1 stock. Like, if all systems pile into AAPL Long, that might be pretty bad. Another down side is I could see it being a pain to update going forward. -Paul --- In amibroker@yahoogroups.com, Gonzaga gonzagags@ wrote: Hi. I am lately trying to mix several systems in on meta-system, and I am observing that is not difficult to obtain good CAR's with low Draw Downs. For example, system 1 trade against 100 tickers of the NAsdaq-100 and system 2, against the same 100 tickers. Both systems 'compete' for the money. This is a 'Meta-system', multi-system and multi-stock. Well, I see it's not very difficult to obtain profitable systems.. I see also that a good filter to improve results is to filter every system with a volatility value of the index you are using, for example ATR of NQ, or ATR of SPX. So you trade any system in the best moment for the system. You have to filter all your systems, and then, mix them in one meta-system. I thing it's not very hard to obtain annualized CARs 30% and DD less than 20%. BUT, it's hard to programme and backtest. Amibroker backtests very easily many stocks, but to mix several systems is a mess.. difficult and easy to fail.. Does anybody know a trading platform that creates this kind of meta-system easily? (perhaps trade station?) Or a way to mix and backtest 2 or 3 systems easily in Ami? Thanks
RE: [amibroker] Re: How to mix systems
Que tal gonzaga? After sending this email, I figured that this might happen: I was answering to Paul's post. Strange Yahoo, sorry for the confusion. I'm afraid I cannot help you on your custom backtest proc, yet. Hi System Mixers, I've been following this thread. I see people want to do different things when they say mix systems. If all you want to do is backtest on daily bars with multiple systems, yet be able to test each system by itself first, one simple way might be to obtain each system's equity curve (preferably walk-forward OOS curve), and then export that curve to a CSV file. Once you have say, 2 or more systems that you want to time / combine, you can then create a new AB database with nothing in it, and then import the equity curves as pseudo/fake symbols using just Date and Close, where the Close is your equity value from that system. Then you could write one code to rank or time the equity curves. Sure, that's quite a bit different than controlling and combining all the individual orders of all the systems, but if you want to keep your system's separate and just time their individual equity curves, then this sort of approach might work. The down side is that you wouldn't be able to see how much risk you are taking on 1 stock. Like, if all systems pile into AAPL Long, that might be pretty bad. Another down side is I could see it being a pain to update going forward. -Paul Answer: As you can see from my questions/answers, I'm always interested in finding new solutions. but quote me if I'm wrong: with your approach you are NOT using ONE equity pool. Say you have 100k$, with your methodology you would do one of the following: Assign 50k$ or 50% to System1 Assign 50k$ or 50% to System2 == Backtest and combine equity curves as you said. Alternativley you might prop up the result in using 100k$ to backtest your 2 Systems. If you operate on a non-percent-risk model, it might work, but a non-percent-risk model ( e.g. fixed number of shares) is not really state-of-art. ...just thinking out loud...
[amibroker] Re: How to mix systems
Thanks my Mike for posting the links. Btw I also contacted support for this issue because if this one proves to be feasible, there will be sort of a lot of money at risk AFTER DEMO-ING the system. I need to be sure. Many thanks also to Marcin from the Amibroker support team that offers sophisticated and fast support. From a theoretical point of view, I came to a conclusion, but please do not ask me about the practical deatils yet, I am not even sure if this is a reliable concept. (Reliable is the keyword in that sentence, I doubt that this concept would be reliable when applied to 3 or even 10 systems) I do NOT require anyone to work out a custom tailored solution for this problem, and I am aware of the AT forum and I read up on autotrading in the user knowledge base. Yes, I also did study one post referring to static variables. As matter of fact, this is what I figured out so far: Mike's post does not work in AT/RT environment. You need to write one MASTER AFL that references/includes all the individual system logics. To avoid overlapping or referencing buy1 to sell2, or sell3, etc. one needs to use STATIC VARIABLES. == REAL TIME TRADING multiple systems is vastly different from BACKTESTING multiple systems. STATIC VARABLES is the concept that I did not understand initially but I read up on it again in greater detail. Another problem with my thought process was that I mixed up BACKTESTING MULTIPLE SYSTEMS and REAL TIME TRADING SYSTEMS. == Mike's Custom Backtest Procedure works on BACKTESTS: NOT AT == Keith's post theoretically suggests the right way for REAL TIME TRADING, NOT BACKTESTING: this approach has been offered from Amibroker support, too (I'll explain later on) What to do: REMEMBER: MULTIPLE SYSTEMS ON THE VERY SAME UNDERLYING You process orders internally to an, let me call it position pool: bar1: SYSTEM A: +2 contracts (2 contracts long) SYSTEMB: -1 contract (1 contract short) total: +1 contract. place order (LONG) for 1 contract. Bar2: SYSTEM A: no signal SYSTEM B: -2 contract place order (SELL) -2 contract overall = -1contract == Amibroker is capable of handling this problem, so to say transmits only the netted/ summed up positions to IB with the restriction of being net flat: [-1 contract ] + [+1 contract] == to Amibroker it means flat. == To me it means what's written there: +1 and -1 I am gonna stop at this point. Thank you all for your kind suggestions and help. Always good to get multiple inputs. This piece of code is gonna be very very very long. Have a nice weekend! Greets from Germany, Farewell for me. Matthias --- In amibroker@yahoogroups.com, Mike sfclimb...@... wrote: What I described was aimed at backtesting multiple systems to emulate how you trade in real life (i.e. generate statistics for how you would have traded, as opposed to what you should trade next). I suspect that the approach that I outlined would not help generate trades for you in real time since by definition the backtest acts on bars that have already closed. I don't auto trade, so I'll leave that advice for others to offer. However, I can point you to some links that can get you started. Your first point of research should probably be the User's Knowledge Base. Herman and Al have written extensively there: http://www.amibroker.org/userkb/ Next, there is a dedicated Yahoo group for AmiBroker auto trading: http://finance.groups.yahoo.com/group/AmiBroker-at/?yguid=83897329 A member of that group, Barry, has posted a skeleton auto trading system to the AFL Library which may be useful: http://www.amibroker.com/members/library/detail.php?id=1225 Mike --- In amibroker@yahoogroups.com, Matthias meridian202@ wrote: Keith and Mike, thank you very much for your helpful postings. Looks like I'll have a long weekend :). Obviously I need to absorb and test what you've just said in great detail which will require time. I shall keep you posted. Just wondering if this is such an akward out-of-the-world problem. Things start to get complex here real quick (for me at least) and whatever approach I take, it seems to be error-prone if not done/checked ultra-carefully. Maybe this one could be implemented a little easier in future versions of Ami so that one could spend more time on system development or portfolio composition rather than spending hours n days to tailor simple AFL's into 1 complex multi-AFL-combined-and-custom-backtest-proc-afl Regarding Keith's post e): // I don't know if AB can properly handle long and short positions simultaneously in the same stock or contract. That will require a simple experiment on your part. If AB can't, then simply export your symbol of choice, import into a spread sheet or text editor, change the symbol to something unique like symbol_SO, and import back into your data base
[amibroker] Re: How to mix systems
Hello, I am having trouble understanding the same issue. If anyone here has a solution I'd like to read up on it as well. Angelo, I have read literally every discussion in this forum regarding this topic and I understand why your reffering to TJ, and please: Yes, I did read the user manual and yes, I did read custom backtester procedure. I believe that I am not the only one who's facing this. Help appreciated. I'd like to post 3 sample strategies for better understanding (which won't work, btw). BASE-TIME: 5min; [ONLY 3 Strategies] Setpositionsize(number, spsshares) // number varies depending on which strategy gets the signal Setoption(maxopenpositions,3) //i.e. every strategy can open 1 position at a time which is not reflected in this command / MA-Crossover // Ema1=ema(c,10); Ema2=ema(c,20); Buy1=cross(ema1,ema2); Sell1=cross(ema2,ema1); Short1=sell1; Cover1=buy1; /// Countertrend/ BBupperiod1= 20; BBlowperiod1= 20; BBupSTD1= 2; BBlowSTD1= 2; BBtop1= BBandTop(C,BBupperiod1,BBupSTD1); BBbot1= BBandBot(C,BBlowperiod1,BBlowSTD1); MA1= ma(C,20); Buy2= Cbbandbot1; Sell2= cma1; Short2= Cbbandtop1; Cover2= Cma1; Buy2= exrem(Buy2,sell2); Sell2= exrem(sell2, buy2); Short2= exrem(short2,cover2); Cover2= exrem(cover2,short2); /// HH LL // TimeFrameSet(inhourly); period1high=20; period1low=20; nhigh=HHV(H,period1high); nlow=LLV(L,period1low); TimeFrameRestore(); Buy3= timeframeexpand(ndayhigh,inhourly) timeframeexpand(ref(ndayhigh,-1), inhourly); Sell3= timeframeexpand(ndaylow,inhourly) timeframeexpand(ref(ndaylow,-1), inhourly); Short3=sell3; Cover3=buy3; REALIME TRADING: What do I need to do is to put these 3 into one AFL for real-time trading - that's what people were saying. How do I arrange the code? I do not trade a lot of underlyings, but say, this one needs to be run on the FDAX(German bluechip future) and on the FDAX only, allowing shorts and longs at the same time. I have read up on the if ( name==) idea. but I want to realtime trade this on ONE underlying only (Trying to express: I cannot change the name of the underlying in realtime trading, so that amibroker would think it's 3 differnt tickers, even though it's only one) Another suggestion was something with static variables, but I didn't grasp the concept. Another suggestion was from Keith Mccombs, suggestion one MASTER-AFL, to control the other afl's. #include afl1 #include afl2 === how do I match buy1 and sell2? BACKTESTING I believe that the issue above differs from the actual backtest-logic and has been vastly mixed causing a lot of confusion. The way I currently see it is, in order to get a proper backtest of multiple systems, one would either need custom backtest proc or --- in this (my case above) case with only one underlying traded --- name the same underlying differently e.g. FDAX1 for system #1 FDAX2 for system #2. Not so elegant, but if it works I'm fine with it. MONEY MANAGEMENT AND PORTFOLIO COMPOSITION Obviously, if you have figured out what and how much to trade, this one is already resolved. But maybe sometimes one would like to try different things or at least check different inputs. Such as: One Equity Pool Several equity pools, manipulate position size if one system is performing poorly, have a system ranking funtion (which system to trade now - based on equity curve, e.g.) do some sort of modern portfolio theory. I don't want to go in the details on MONEY MANAGEMENT AND PORTFOLIO COMPOSITION and for now I'd be delighted if someone could show me how to match my buy/sell signals for realtime trading. One software which I know of and is capable of adressing all these problems easier is rina portfolio maestro. http://www.portfoliomaestro.com/ I haven't tried it and I'm also not planning on doing so (10k$ per year), but I wanted to understand what's possible. Greetings, M --- In amibroker@yahoogroups.com, Jeff jeffro...@... wrote: In any case you'll have to decide on the logic to decide on allocation. After you do that, You could make the meta system a linear combination of the two. Create multiple randoms runs of each and assign a static variable, to the random equity curves. That way you could do a Monte Carlo analysis with both sets together and different allocations. But then again I would I strongly suggest researxhing and understanding the characters of the tendencies you are exploiting instead of smashing them together blindly. --- In amibroker@yahoogroups.com, Gonzaga gonzagags@ wrote: Hi. I am lately trying to mix several systems in on meta-system, and I am observing that is not difficult to obtain good CAR's with low Draw Downs. For example, system 1 trade against 100 tickers of the NAsdaq-100 and
[amibroker] Re: How to mix systems
Sorry for the typo in the Backtesting-Paragraph: Should read: match buy1 and sell1 (NOT: buy1 and sell2) --- In amibroker@yahoogroups.com, Matthias meridian...@... wrote: Hello, I am having trouble understanding the same issue. If anyone here has a solution I'd like to read up on it as well. Angelo, I have read literally every discussion in this forum regarding this topic and I understand why your reffering to TJ, and please: Yes, I did read the user manual and yes, I did read custom backtester procedure. I believe that I am not the only one who's facing this. Help appreciated. I'd like to post 3 sample strategies for better understanding (which won't work, btw). BASE-TIME: 5min; [ONLY 3 Strategies] Setpositionsize(number, spsshares) // number varies depending on which strategy gets the signal Setoption(maxopenpositions,3) //i.e. every strategy can open 1 position at a time which is not reflected in this command / MA-Crossover// Ema1=ema(c,10); Ema2=ema(c,20); Buy1=cross(ema1,ema2); Sell1=cross(ema2,ema1); Short1=sell1; Cover1=buy1; /// Countertrend / BBupperiod1= 20; BBlowperiod1= 20; BBupSTD1= 2; BBlowSTD1= 2; BBtop1= BBandTop(C,BBupperiod1,BBupSTD1); BBbot1= BBandBot(C,BBlowperiod1,BBlowSTD1); MA1= ma(C,20); Buy2= Cbbandbot1; Sell2= cma1; Short2= Cbbandtop1; Cover2= Cma1; Buy2= exrem(Buy2,sell2); Sell2= exrem(sell2, buy2); Short2= exrem(short2,cover2); Cover2= exrem(cover2,short2); /// HH LL // TimeFrameSet(inhourly); period1high=20; period1low=20; nhigh=HHV(H,period1high); nlow=LLV(L,period1low); TimeFrameRestore(); Buy3= timeframeexpand(ndayhigh,inhourly) timeframeexpand(ref(ndayhigh,-1), inhourly); Sell3= timeframeexpand(ndaylow,inhourly) timeframeexpand(ref(ndaylow,-1), inhourly); Short3=sell3; Cover3=buy3; REALIME TRADING: What do I need to do is to put these 3 into one AFL for real-time trading - that's what people were saying. How do I arrange the code? I do not trade a lot of underlyings, but say, this one needs to be run on the FDAX(German bluechip future) and on the FDAX only, allowing shorts and longs at the same time. I have read up on the if ( name==) idea. but I want to realtime trade this on ONE underlying only (Trying to express: I cannot change the name of the underlying in realtime trading, so that amibroker would think it's 3 differnt tickers, even though it's only one) Another suggestion was something with static variables, but I didn't grasp the concept. Another suggestion was from Keith Mccombs, suggestion one MASTER-AFL, to control the other afl's. #include afl1 #include afl2 === how do I match buy1 and sell2? BACKTESTING I believe that the issue above differs from the actual backtest-logic and has been vastly mixed causing a lot of confusion. The way I currently see it is, in order to get a proper backtest of multiple systems, one would either need custom backtest proc or --- in this (my case above) case with only one underlying traded --- name the same underlying differently e.g. FDAX1 for system #1 FDAX2 for system #2. Not so elegant, but if it works I'm fine with it. MONEY MANAGEMENT AND PORTFOLIO COMPOSITION Obviously, if you have figured out what and how much to trade, this one is already resolved. But maybe sometimes one would like to try different things or at least check different inputs. Such as: One Equity Pool Several equity pools, manipulate position size if one system is performing poorly, have a system ranking funtion (which system to trade now - based on equity curve, e.g.) do some sort of modern portfolio theory. I don't want to go in the details on MONEY MANAGEMENT AND PORTFOLIO COMPOSITION and for now I'd be delighted if someone could show me how to match my buy/sell signals for realtime trading. One software which I know of and is capable of adressing all these problems easier is rina portfolio maestro. http://www.portfoliomaestro.com/ I haven't tried it and I'm also not planning on doing so (10k$ per year), but I wanted to understand what's possible. Greetings, M --- In amibroker@yahoogroups.com, Jeff jeffro861@ wrote: In any case you'll have to decide on the logic to decide on allocation. After you do that, You could make the meta system a linear combination of the two. Create multiple randoms runs of each and assign a static variable, to the random equity curves. That way you could do a Monte Carlo analysis with both sets together and different allocations. But then again I would I strongly suggest researxhing and understanding the characters of the tendencies you are exploiting
[amibroker] Re: How to mix systems
Thanks Keith for joining this discussing. I don't quite know who you're trying to adress with your question/suggestion. Assuming you refer to my previous post: 1. Put those three Systems in one afl as you would when autotrading: Each System has its BUY and SELL and SHORT and COVER: I want system1 to refer only and only to its predetermined buy1,sell1,short1,cover1 rules 2. Each System uses ATR positionsizing, obviously not the same number of contracts traded due to different timeframes and different system logic 3. FDAX is a futures contract: The systems are intended to share the same equity pool [to be precise: closed equity] 4. All Systems trade this ONE underlying only 5. Each System is allowed to open ONE position at a time, not matter what positions of the remaining systems are open or closed 6. As stated in the text below: Same time longs and shorts of different systems are allowed: E.g. system1(long), system2(short), system3(no-pos) 7. Each system has its own predetermined exits, no matter what the other systems do [edit: in order to NOT complicate things further, I skipped Applystop in the system logic of my example in the previous post] 8. Backtest each system individually is no prob, together with FDAX1,FDAX2 (faking different underlyings to amibroker)... no prob either So: The way I understand it currently is: For Ami to autotrade, I need to write the logic of the 3 systems into 1 AFL. Taking all the above mentioned points into consideration: How does one do that? In that sense you are right: Possibly Mixing Systems as a topic is a bit weak in determining what each user wants. My idea was to breakdown this whole topic in 3 main areas: *Realtime-Trading environment and relating issues *Backtesting multiple Systems simultaneously in general *Broken down further in various portfolio and moneymangement techniques: which MM and Portfoilio Management is best for my Trading system basket, individual system ranking (with equity curve, don't know a better metric for the system than equity curve), or also the other topics involved: Walk-Forward Efficiency Ratio (for each individual system), Modern Portfolio Design (here referring to: Multiple Systems logics on different underlying to produce smoothest, risk-adjusted PORTFOLIO equity) Hope I made myself clearer, but you're right the topic is huge and some of it has been discussed some time ago. The thing that gives me a headache is that I couldn't find a practicle example that would gather together the conclusions out of the discussion. The idea to break this topic down into the 3 parts are a suggestion to make things easier, I'm not sure if it's the best solution due to overlapping. But come on, we shot a rocket to the moon, this little there needs to be possible. Thanks, M --- In amibroker@yahoogroups.com, Keith McCombs kmcco...@... wrote: I've read a number of posts on the thread. And it seems to me that not everyone has the same meaning in mind when they are talking about a 'MIX' system. What do you mean by a mix system? I assume that two or more systems are supposed to share SOMETHING. Is that SOMETHING, 1. The same equities? If so, what do you want to happen if two or more systems want to go in different directions, perhaps one holds while another sells, or even another shorts? Once you have answered these questions, you might find that sigScaleIn and sigScaleOut will help you out. 2. Available cash? If so, to what extent? If you know, then maybe use setPositionSize() and/or PositionScale can help you. 3. Risk and rewards, but keeping money from profits and losses separate? If so, try backtesting separately. Then save ~~~Equity results with different names and add them in AB, with buy and hold for each. 4. Or, maybe something else? -- Keith On 7/1/2010 17:29, Matthias wrote: Sorry for the typo in the Backtesting-Paragraph: Should read: match buy1 and sell1 (NOT: buy1 and sell2) --- In amibroker@yahoogroups.com mailto:amibroker%40yahoogroups.com, Matthias meridian202@ wrote: Hello, I am having trouble understanding the same issue. If anyone here has a solution I'd like to read up on it as well. Angelo, I have read literally every discussion in this forum regarding this topic and I understand why your reffering to TJ, and please: Yes, I did read the user manual and yes, I did read custom backtester procedure. I believe that I am not the only one who's facing this. Help appreciated. I'd like to post 3 sample strategies for better understanding (which won't work, btw). BASE-TIME: 5min; [ONLY 3 Strategies] Setpositionsize(number, spsshares) // number varies depending on which strategy gets the signal Setoption(maxopenpositions,3) //i.e. every strategy can open 1 position at a time which is not reflected in this command
[amibroker] Re: How to mix systems
Keith and Mike, thank you very much for your helpful postings. Looks like I'll have a long weekend :). Obviously I need to absorb and test what you've just said in great detail which will require time. I shall keep you posted. Just wondering if this is such an akward out-of-the-world problem. Things start to get complex here real quick (for me at least) and whatever approach I take, it seems to be error-prone if not done/checked ultra-carefully. Maybe this one could be implemented a little easier in future versions of Ami so that one could spend more time on system development or portfolio composition rather than spending hours n days to tailor simple AFL's into 1 complex multi-AFL-combined-and-custom-backtest-proc-afl Regarding Keith's post e): // I don't know if AB can properly handle long and short positions simultaneously in the same stock or contract. That will require a simple experiment on your part. If AB can't, then simply export your symbol of choice, import into a spread sheet or text editor, change the symbol to something unique like symbol_SO, and import back into your data base. Now you have two separate symbols, which we know AB can handle properly. /// Just by reading it, I figured that one would require an external source, such as Excel. Problem here is that this is supposed to work in RT environment doing autotrading, anyways I'll look into it. Regarding Mike's post: /// I haven't actually tried any of the above, and do not know how well it would translate into auto trading /// Sounds like this is something extraordinary, but are there so little user asking about this multiple-system concept? Look, I don't have an account to autotrade yet, but if I figured out a way on how to do this, this would be run in AA window, backtesting the system, so to say - run that piece of code every xx minutes, is that correct? I remember somebody from Amibroker support told me to program it like a scan and think of autotrading like I would just scan for signals. Not sure if I understand the concept, but does scan include custom backtest proc? I'll see what goes, but out curiosity: just wondering if the execution speed of this code would be proefficient. Thanks again.
[amibroker] Re: inconsistent EOD data
There is literally NO workaround to the problem you just described. I agree, due to different timestamps of data, especially with finam (Russia) these differences can be explained. But this issue becomes more and more unpredictable the shorter the time frame you are trading, i.e. one minuter or one day. 10 pips on a daily move just don't mean that much as on a 1 minute chart especially on news releases or at important support/resistance levels. But it's already been mentioned: go for futures at the CME, this market is regulated whereas the brokers simply are not. Or to put it in different words: FX is an unregulated market where banks can make up the price they want within certain limits (especially at an accumulation of stops). Something more to worry: http://www.forexfactory.com/showthread.php?t=70582 Well that's kinda nasty, but it's true when dealing forex. In his book Beat the Forex Dealer by Augustin Silvani, he even explains that a well known FX-Broker, called FXC(guess the last letter) state that these folks are big on advertising no-dealing-desk-execution whereas no dealing desk only means that the desk has been replaced with a sophisticated algorhitm that's run by a computer. Suggestion: Trade breakouts in the direction of the breakout on shorter time frame or benefit from the longer term trend. PS Finam uses indicative data, thus it's sort of a extra-multi-bank composition. If you use other brokers, they might get their quotes from a more limited number of source banks. In FX: the only quote that counts is the quote of your broker, so be wary when buying data from different vendors other than the broker you are dealing with. Regards, M. --- In amibroker@yahoogroups.com, rutle1 rut...@... wrote: Consider that we went to Daylight Savings Time here in the EST, and Russian time (finam.com) didn't change yet? So maybe the quotes are 1 hour off until their timezone changes? --- In amibroker@yahoogroups.com, Bob Pitatzidis bobz2335@ wrote: Hello all I just checked the End of Day price for forex-EURUSD in 3 different software. They all report different prices: April 8th: VTTrader2: 1,34929 Metarader5: 1,34984 Amibroker (from finam.com): 1,3626 April 9th: VTTrader2: 1,33595 Metarader5: 1,33572 Amibroker (from finam.com): 1,33480 I thought that it could be some hour difference, ie GMT vs EST or something like this. BUT April 9th is end of day AND end of the week. They should all report the same price. Any ideas? What site/source can I check for reliable price? Thanks Bob __ Information from ESET Smart Security, version of virus signature database 5018 (20100411) __ The message was checked by ESET Smart Security. http://www.eset.com
[amibroker] Re: Different Applystops for long and short positions possible?
Hi, I use this one: buypos=Flip(Buy,Sell); Shortpos=Flip(Short,Cover); SLbuy=Optimize(SLbuy,0.5,0.3,0.7,0.05); If you wanna optimize your stops and profit targets SLshort=Optimize(SLshort,0.5,0.3,0.7,0.05); TPbuy=Optimize(TPbuy,1,0.9,1.2,0.05); TPshort=Optimize(TPshort,1,0.9,1.2,0.05); SL=IIf(Buypos,SLbuy,IIf(Shortpos,slshort,0)); TP=IIf(Buypos,TPbuy,IIf(Shortpos,tpshort,0)); ApplyStop(stopTypeLoss,stopModePercent,SL); ApplyStop(stopTypeProfit,stopModePercent,TP); Sorry for the strange line breaks... Cheers, Matthias --- In amibroker@yahoogroups.com, Markus Witzler funny...@... wrote: Hello, I want to specify different appystop conditions for long and short posisions. How does AB recognize which one is for which trade? Thanks Markus __ Information from ESET Smart Security, version of virus signature database 4668 (20091207) __ The message was checked by ESET Smart Security. http://www.eset.com
[amibroker] Re:
Hi, Mike, thanks for helping me out on this. Your code was completely right, I made a mistake counting the crossovers, I tried it with other symbols. I appreciated your help, thank you very much again, even though my questions were starter-questions. Matthias
[amibroker] Re: Could someone comment on how this loop is executed (Basic)
--- In amibroker@yahoogroups.com, Mike sfclimb...@... wrote:
Hi,
Nothing wrong with wanting to understand the mechanics. Before getting
into that, however, I did not understand your comment
it starts to count wrong(should count to 3, but counts to 2 only
considering the 1st triple of crossovers, probably due to initializing
the EMA200)
What exactly do you mean by that? Can you give a specific example citing
a specific symbol over a specific date range and indicate what you
expected to see vs. what you are seeing?
Looking at your original post;- Your first attempt should not be using
AND. Two separate statements separated by ';' is the correct way to go,
as you had done in the second attempt.
- In your second attempt, you are using the wrong condition for
incrementing Counter. You should be checking for true values (i.e. non
zero).
Here is a corrected version of your code. I believe that it will give
you what you want, and, in my tests anyway, line up exactly with the non
looping version that I sent you originally.
/* Option 1
EMA1 = EMA( C, 10 );
EMA2 = EMA( C, 200 );
Plot( EMA1, EMA10, colorRed, styleLine );
Plot( EMA2, EMA20, colorBlue, styleLine );
Buycross[0] = 0;
exit[0] = 0;
threshold = 3;
multiplier = 1;
Buycross = Cross( EMA1, EMA2 );
Counter = Cum( Buycross );
// start loop
for ( i = 0; i BarCount;i++ )
{
if ( Counter[i] == threshold*multiplier )
{
multiplier++;
exit[i] = 1; // separate statements, not AND!
}
}
printf( buycross:\n );
WriteVal( Buycross, 1 );
printf( counter:\n );
WriteVal( Counter, 1 );
printf( multiplier:\n );
WriteVal( multiplier, 1 );
printf( exit:\n );
WriteVal( exit, 1 );
*/
/* Option 2
EMA1 = EMA( C, 10 );
EMA2 = EMA( C, 200 );
Plot( EMA1, EMA10, colorRed, styleLine );
Plot( EMA2, EMA20, colorBlue, styleLine );
Buycross[0] = 0;
exit[0] = 0;
COUNTER = 0;
threshold = 3;
multiplier = 1;
Buycross = Cross( EMA1, EMA2 );
// start loop
for ( i = 0; i BarCount;i++ )
{
if ( buycross[i] != 0 ) // Check for NOT zero!
{
counter++;
}
if ( Counter == threshold*multiplier )
{
multiplier++;
exit[i] = 1;
}
}
printf( buycross:\n );
WriteVal( Buycross, 1 );
printf( counter:\n );
WriteVal( Counter, 1 );
printf( multiplier:\n );
WriteVal( multiplier, 1 );
printf( exit:\n );
WriteVal( exit, 1 );
*/
Mike
--- In amibroker@yahoogroups.com, Matthias m.kuschel31@ wrote:
Hello,
Thanks to all for the prompt reply. (I just figured out how to format
this, my 1st post looked a little strange).
EMA1 = EMA(C, 10);
EMA2 = EMA(C, 200);
Buycross = Cross(EMA1, EMA2);
Counter = Cum(Buycross);
Triples = Cross(Counter % 3 == 0, 0);
Plot(EMA1,EMA10, colorRed,styleLine);
Plot(EMA2,EMA20,colorBlue,styleLine);
Plot(Triples, Triples, colorGreen, styleOwnScale | styleHistogram);
This code works absolutely fine, many thanks, only problem here: it
starts to count wrong(should count to 3, but counts to 2 only
considering the 1st triplele of crossovers, probably due to
intializing
the EMA200) any suggestions? But I need you to understand where I am
coming from:
Obviously you are much more experienced in writing afl than me and I
get
Mike's basic concept of avoiding loops for proefficient coding and the
EMA-plot-labeling was not correct.
You have to understand that I am trying to realize/code my ideas and
the
2nd code is where I started off and probably would start off again, be
it good or bad programming. I do have your solution to my problem now,
but I still don't understand why my concept does not do the job. Of
course I did read the articles on How AFL works, this was were I
started and also got a hand on the barcount funtion which I thought
was
the problem (not counting the whole OHLC-array due to faster
processing
speed). I also asked my brother on this simple one, who is used to
programming C. As AFL is C-based I would have assumed this code
should
work. Please help me to understand what went wrong / is wrong in my
logic here, I consider understanding this basic concept to be very
important for future purpose.
Thanks in advance,
Matthias
Hi,
Mike, thanks for helping me out on this. Your code was completely right, I made
a mistake counting the crossovers, I tried it with other symbols. I appreciated
your help, thank you very much again, even though my questions were
starter-questions .
Matthias
[amibroker] Re: Could someone comment on how this loop is executed (Basic)
Hello, Thanks to all for the prompt reply. (I just figured out how to format this, my 1st post looked a little strange). EMA1 = EMA(C, 10); EMA2 = EMA(C, 200); Buycross = Cross(EMA1, EMA2); Counter = Cum(Buycross); Triples = Cross(Counter % 3 == 0, 0); Plot(EMA1,EMA10, colorRed,styleLine); Plot(EMA2,EMA20,colorBlue,styleLine); Plot(Triples, Triples, colorGreen, styleOwnScale | styleHistogram); This code works absolutely fine, many thanks, only problem here: it starts to count wrong(should count to 3, but counts to 2 only considering the 1st triplele of crossovers, probably due to intializing the EMA200) any suggestions? But I need you to understand where I am coming from: Obviously you are much more experienced in writing afl than me and I get Mike's basic concept of avoiding loops for proefficient coding and the EMA-plot-labeling was not correct. You have to understand that I am trying to realize/code my ideas and the 2nd code is where I started off and probably would start off again, be it good or bad programming. I do have your solution to my problem now, but I still don't understand why my concept does not do the job. Of course I did read the articles on How AFL works, this was were I started and also got a hand on the barcount funtion which I thought was the problem (not counting the whole OHLC-array due to faster processing speed). I also asked my brother on this simple one, who is used to programming C. As AFL is C-based I would have assumed this code should work. Please help me to understand what went wrong / is wrong in my logic here, I consider understanding this basic concept to be very important for future purpose. Thanks in advance, Matthias
