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

RE: [amibroker] Optimizing for consistent performance



PureBytes Links

Trading Reference Links


What 
you need to do is "find a way" to calculate your profits in a time segemented 
way. Ie. Theoretical profits for each week or month.  Then plot that on 
Herman's chart for smoothness.  You'll probably have to do this away from 
AB. You'll need end of week pricing along with your positions as determinedin 
AB. If you put those 2 items together in a spreadsheet you should be able to 
calc the theoretical profits at the end of each week.  You'll need to do 
this via a macro.
<FONT face=Arial color=#0000ff 
size=2> 
If you 
happend to be using TC2000 or Quotes Plus then you can access the pricing via an 
Excel macro and use that plus the exported trade list to come up with 
something...
<FONT face=Arial color=#0000ff 
size=2> 
Just 
thinking out loud on this ...
<FONT face=Arial color=#0000ff 
size=2> 
<FONT face=Arial color=#0000ff 
size=2>dingo


<FONT 
face=Tahoma size=2>-----Original Message-----From: Steve Davis 
[mailto:sdavis@xxxx] Sent: Thursday, October 17, 20022:45 
PMTo: amibroker@xxxxxxxxxxxxxxxSubject: RE: [amibroker] 
Optimizing for consistent performance
<SPAN 
class=522583218-17102002>Rick,
<SPAN 
class=522583218-17102002> 
<SPAN 
class=522583218-17102002>Hermans 3D optimizer is a blast to play with. I am 
having great fun exploring new ideas. You make a g<FONT 
face=Arial color=#0000ff size=2>ood point, 
sharp peaks bad, rolling hills good. I will look for that.
<SPAN 
class=522583218-17102002> 
I'm 
not sure if I'm adequately expressing my original point. 
<SPAN 
class=522583218-17102002>The z-axis of the 3D plot is total system return over 
the 50-year test. I am suggesting perhaps there is some other system 
measurement that could be used for the z-axis. I need a mathematical 
expression that evaluates the return of the system but also takes into account 
the volatility of the equity curve.
<SPAN 
class=522583218-17102002> 
<SPAN 
class=522583218-17102002>Thanks for the discussion. Putting this into words 
helps me clarify my own thoughts.
<SPAN 
class=522583218-17102002>-Steve
<SPAN 
class=522583218-17102002> 

<FONT face=Tahoma 
size=2>-----Original Message-----From: Rick Parsons 
[mailto:RickParsons@xxxx]Sent: Thursday, October 17, 
2002 2:23 PMTo: amibroker@xxxxxxxxxxxxxxxSubject:RE: 
[amibroker] Optimizing for consistent performance
<FONT color=#000080 
size=2>Steve,
<FONT color=#000080 
size=2> 
Re: 
>>it is designed to optimize the 
parameters to maximize the total system 
return.<<
<FONT color=#000080 
size=2> 
Just my 
humble opinion but I suspect you are using the 3-D for the wrong 
purposes.  If you want maximum returns, you can simply optimize inAA 
and sort by % return.
<FONT color=#000080 
size=2> 
3-D is to 
show you which parameters are the most robust.  If you find a large 
flat plain in the 3-D that means there is a wide range of parameters that 
give consistent returns.  This is robustness.
<FONT color=#000080 
size=2> 
Ifyou use 
3-D to find the best return, chances are it is located on a peak with sharp 
drop-offs (although this is not always the case).  The sharp drop-offs 
indicate that the parameters are not robust.
<FONT color=#000080 
size=2> 
Tosome 
extent, robustness helps with "consistent performance".
<FONT color=#000080 
size=2> 
<FONT face="Vladimir Script" color=#000080 
size=5>Rick

<FONT face=Tahoma 
size=2>-----Original Message-----From: Steve Davis 
[mailto:sdavis@xxxx]Sent: Thursday, October 17, 2002 
2:07 PMTo: amibroker@xxxxxxxxxxxxxxxSubject: RE: 
[amibroker] Optimizing for consistent performance
<SPAN 
class=594000118-17102002>Rick,
<SPAN 
class=594000118-17102002> 
<SPAN 
class=594000118-17102002>Yes, I have tried Herman's optimizer -- itis 
designed to optimze the parameters to maximize the total system return. 
That is useful, but not exactly what I want. If I optimize the systemto 
maximize gains, the drawdowns become horrible. I need a system that 
produces steady monthly returns.
<SPAN 
class=594000118-17102002> 
<SPAN 
class=594000118-17102002>You can get at least 50 years of historical EOD 
data from Yahoo. The DOW, SP500, Utilities, and transports all have at 
least 50 years of data available.
<SPAN 
class=594000118-17102002> 
<SPAN 
class=594000118-17102002>-Steve

<FONT face=Tahoma 
size=2>-----Original Message-----From: Rick Parsons 
[mailto:RickParsons@xxxx]Sent: Thursday, October 
17, 2002 2:01 PMTo: 
amibroker@xxxxxxxxxxxxxxxSubject: RE: [amibroker] Optimizing 
for consistent performance
<FONT color=#000080 
size=2>Steve,
Not 
exactly sure what you are asking.  But have you tried Herman's3-D 
Optimizer for Excel (in download area, I believe) to find the most 
robust parameters?
<FONT color=#000080 
size=2> 
Another 
question:  Where did you get 50 years of data?
<FONT color=#000080 
size=2> 
<FONT color=#000080 
size=2>Thanks,
 
<FONT face="Vladimir Script" color=#000080 
size=5>Rick

<FONT face=Tahoma 
size=2>-----Original Message-----From: Steve Davis 
[mailto:sdavis@xxxx]Sent: Thursday, October 17, 
2002 1:34 PMTo: 
amibroker@xxxxxxxxxxxxxxxSubject: [amibroker] Optimizing 
for consistent performance
AB rocks! 
<SPAN 
class=437573315-17102002>After using AB for only a month, I have 
developed a system that produces 50% 
risk-adjusted-returns trading the S&P500 over the last 
50 years. It is a short-term trading system with an average trade 
lasting 4 days.
<SPAN 
class=437573315-17102002> 
But there 
is some spikiness in the equity curve. I am having some difficulty 
tuning the system to produce steady monthly cash flow. The goal is not 
to maximize profits over 50 years, but to produce consistent 
profitable results.
<SPAN 
class=437573315-17102002> 
One 
problem is how to measure the system performance so it can be properly 
tuned. I do not know many important statistics of the system. For 
example, has the system ever had a losing year? What is the worst-case 
number of consecutive losing months? <FONT face=Arial 
size=2>I don't think minimizing 
drawdown is sufficient to meet my goals.
<SPAN 
class=437573315-17102002> 
The 
approach I have been using is an iterative process of eye-balling 
the equity curve, changing parameters, and trying again. There must be 
a better way. Is there a reasonable way to automate this 
task?
<SPAN 
class=437573315-17102002> 
How are 
other AB users optimizing their equity curve?
 
<SPAN 
class=437573315-17102002>Cheers,
<SPAN 
class=437573315-17102002>-StevePost 
AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web 
page: <A 
href="">http://groups.yahoo.com/group/amiquote/messages/)Check 
group FAQ at: <A 
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of 
Service. Post AmiQuote-related 
messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A 
href="">http://groups.yahoo.com/group/amiquote/messages/)Check 
group FAQ at: <A 
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of 
Service. Post AmiQuote-related 
messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A 
href="">http://groups.yahoo.com/group/amiquote/messages/)Check 
group FAQ at: <A 
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of Service. 
Post AmiQuote-related messages ONLY to: 
amiquote@xxxxxxxxxxxxxxx (Web page: <A 
href="">http://groups.yahoo.com/group/amiquote/messages/)Check 
group FAQ at: <A 
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of Service. 
Post 
AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A 
href="">http://groups.yahoo.com/group/amiquote/messages/)Check 
group FAQ at: <A 
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of Service.