Evaluating equity curves

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Evaluating equity curves

aleks.clark@gmail.com
As part of a project I'm working on that uses genetic algorithms to
optimize trading parameters, I find myself seeking a way to evaluate
the equity curve that results from a given set of trading rules. It
seems to be an obvious area of research, so I was wondering what's
available in R-land or just in the world of finance in general. I've
poked around with splines as a way to express how 'nice' an equity
curve is (steady upward rise as opposed to a "jagged" line), but I
feel that there are probably better ways to do things...


Thanks,


--
Aleks Clark

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

R. Vince
Do NOT bother with this, Aleks, until you first read about the First and
particularly Second Arc Sine Laws and how they might pertain to equity
curves. (To avoid fitting splinies or anything else to unicorn
costumes) -Ralph Vince
----- Original Message -----
From: "Aleks Clark" <[hidden email]>
To: <[hidden email]>
Sent: Sunday, October 11, 2009 6:47 PM
Subject: [R-SIG-Finance] Evaluating equity curves


> As part of a project I'm working on that uses genetic algorithms to
> optimize trading parameters, I find myself seeking a way to evaluate
> the equity curve that results from a given set of trading rules. It
> seems to be an obvious area of research, so I was wondering what's
> available in R-land or just in the world of finance in general. I've
> poked around with splines as a way to express how 'nice' an equity
> curve is (steady upward rise as opposed to a "jagged" line), but I
> feel that there are probably better ways to do things...
>
>
> Thanks,
>
>
> --
> Aleks Clark
>
> _______________________________________________
> [hidden email] mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only.
> -- If you want to post, subscribe first.

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

braverock
In reply to this post by aleks.clark@gmail.com
Aleks Clark wrote:
> As part of a project I'm working on that uses genetic algorithms to
> optimize trading parameters, I find myself seeking a way to evaluate
> the equity curve that results from a given set of trading rules. It
> seems to be an obvious area of research, so I was wondering what's
> available in R-land or just in the world of finance in general. I've
> poked around with splines as a way to express how 'nice' an equity
> curve is (steady upward rise as opposed to a "jagged" line), but I
> feel that there are probably better ways to do things...

Having worked both in quantitative trading and in more traditional asset
management roles, I've never quite understood the artificial distinction
between "equity curves" and any other kind of returns.  In my experience, all
the usual performance and risk analysis tools (amply provided for in R) as well
as attribution (e.g. Bacon, much of which is implemented in fPortfolio) are
equally applicable to trading strategies as they are to more traditional
investment.  Also see Pat Burns' paper on evaluatinfg trading strategies for
additional ideas.

Regards,

   - Brian

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

R. Vince
Brian,

Isn't any stream of cumulative returns, de facto, an equity curve? Or am I
misunderstanding this? Thanks, Ralph Vince

----- Original Message -----
From: "Brian G. Peterson" <[hidden email]>
To: "Aleks Clark" <[hidden email]>
Cc: <[hidden email]>
Sent: Sunday, October 11, 2009 8:46 PM
Subject: Re: [R-SIG-Finance] Evaluating equity curves


> Aleks Clark wrote:
>> As part of a project I'm working on that uses genetic algorithms to
>> optimize trading parameters, I find myself seeking a way to evaluate
>> the equity curve that results from a given set of trading rules. It
>> seems to be an obvious area of research, so I was wondering what's
>> available in R-land or just in the world of finance in general. I've
>> poked around with splines as a way to express how 'nice' an equity
>> curve is (steady upward rise as opposed to a "jagged" line), but I
>> feel that there are probably better ways to do things...
>
> Having worked both in quantitative trading and in more traditional asset
> management roles, I've never quite understood the artificial distinction
> between "equity curves" and any other kind of returns.  In my experience,
> all the usual performance and risk analysis tools (amply provided for in
> R) as well as attribution (e.g. Bacon, much of which is implemented in
> fPortfolio) are equally applicable to trading strategies as they are to
> more traditional investment.  Also see Pat Burns' paper on evaluatinfg
> trading strategies for additional ideas.
>
> Regards,
>
>   - Brian
>
> _______________________________________________
> [hidden email] mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only.
> -- If you want to post, subscribe first.

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

braverock
R. Vince wrote:
> Brian,
>
> Isn't any stream of cumulative returns, de facto, an equity curve? Or am
> I misunderstanding this? Thanks, Ralph Vince

You're absolutely correct.  This is why I described it as an artificial
distinction.  Many traders seem to believe that there is something magical
about an "equity curve" when it is really just a way of describing cumulative
returns. Thus my point that there is indeed nothing magical, and all the usual
suspects on return, risk, and attribution analysis may be readily applied to
analysis of trading strategies.

Regards,

     - Brian

> ----- Original Message -----
 > From: "Brian G. Peterson" <[hidden email]>

>> Aleks Clark wrote:>>> As part of a project I'm working on that uses genetic algorithms to
>>> optimize trading parameters, I find myself seeking a way to evaluate
>>> the equity curve that results from a given set of trading rules. It
>>> seems to be an obvious area of research, so I was wondering what's
>>> available in R-land or just in the world of finance in general. I've
>>> poked around with splines as a way to express how 'nice' an equity
>>> curve is (steady upward rise as opposed to a "jagged" line), but I
>>> feel that there are probably better ways to do things...
>>
>> Having worked both in quantitative trading and in more traditional
>> asset management roles, I've never quite understood the artificial
>> distinction between "equity curves" and any other kind of returns.  In
>> my experience, all the usual performance and risk analysis tools
>> (amply provided for in R) as well as attribution (e.g. Bacon, much of
>> which is implemented in fPortfolio) are equally applicable to trading
>> strategies as they are to more traditional investment.  Also see Pat
>> Burns' paper on evaluatinfg trading strategies for additional ideas.
>>
>> Regards,
>>
>>   - Brian
>>
>> _______________________________________________
>> [hidden email] mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>> -- Subscriber-posting only.
>> -- If you want to post, subscribe first.


--
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

Robert Sams
In reply to this post by braverock
Quite right, Ralph. More precisely: any series of cumulative returns
plotted against time on a 2d chart is, *by definition*, an equity curve.


Aleks, I'm not sure you're after. To me, the return series of a trading
strategy isn't 'evaluated' per se, it's evaluated with respect to some
specific questions or conjectures you have regarding it. Articulating
this is logically prior to choosing the appropriate mathematical
technique.

If you can be more specific, we might be able to point you in the right
direction.

Robert

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf Of R. Vince
> Sent: 12 October 2009 02:06
> To: Brian G. Peterson; Aleks Clark
> Cc: [hidden email]
> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>
> Brian,
>
> Isn't any stream of cumulative returns, de facto, an equity
> curve? Or am I misunderstanding this? Thanks, Ralph Vince
>
> ----- Original Message -----
> From: "Brian G. Peterson" <[hidden email]>
> To: "Aleks Clark" <[hidden email]>
> Cc: <[hidden email]>
> Sent: Sunday, October 11, 2009 8:46 PM
> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>
>
> > Aleks Clark wrote:
> >> As part of a project I'm working on that uses genetic algorithms to
> >> optimize trading parameters, I find myself seeking a way
> to evaluate
> >> the equity curve that results from a given set of trading rules. It
> >> seems to be an obvious area of research, so I was wondering what's
> >> available in R-land or just in the world of finance in
> general. I've
> >> poked around with splines as a way to express how 'nice' an equity
> >> curve is (steady upward rise as opposed to a "jagged" line), but I
> >> feel that there are probably better ways to do things...
> >
> > Having worked both in quantitative trading and in more
> traditional asset
> > management roles, I've never quite understood the
> artificial distinction
> > between "equity curves" and any other kind of returns.  In
> my experience,
> > all the usual performance and risk analysis tools (amply
> provided for in
> > R) as well as attribution (e.g. Bacon, much of which is
> implemented in
> > fPortfolio) are equally applicable to trading strategies as
> they are to
> > more traditional investment.  Also see Pat Burns' paper on
> evaluatinfg
> > trading strategies for additional ideas.
> >
> > Regards,
> >
> >   - Brian
> >
> > _______________________________________________
> > [hidden email] mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> > -- Subscriber-posting only.
> > -- If you want to post, subscribe first.
>
> _______________________________________________
> [hidden email] mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only.
> -- If you want to post, subscribe first.
>

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

aleks.clark@gmail.com
Well, the reasoning went like this...since I can instantly look at a
equity curve and say 'hey that's a nice steady upward slope' as
opposed to "wow that's a jaggy one with lots of drawdowns", I thought
to apply some sort of maths to turn this gestalt into a number that
could be taken along with the final balance as a measure of fitness,
thus the applying of splines. Now that you mention it however, there
are probably better ways of evaluating these strategies that aren't
'graphical' in nature.

On Mon, Oct 12, 2009 at 6:45 AM, Robert Sams <[hidden email]> wrote:

> Quite right, Ralph. More precisely: any series of cumulative returns
> plotted against time on a 2d chart is, *by definition*, an equity curve.
>
>
> Aleks, I'm not sure you're after. To me, the return series of a trading
> strategy isn't 'evaluated' per se, it's evaluated with respect to some
> specific questions or conjectures you have regarding it. Articulating
> this is logically prior to choosing the appropriate mathematical
> technique.
>
> If you can be more specific, we might be able to point you in the right
> direction.
>
> Robert
>
>> -----Original Message-----
>> From: [hidden email]
>> [mailto:[hidden email]] On Behalf Of R. Vince
>> Sent: 12 October 2009 02:06
>> To: Brian G. Peterson; Aleks Clark
>> Cc: [hidden email]
>> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>>
>> Brian,
>>
>> Isn't any stream of cumulative returns, de facto, an equity
>> curve? Or am I misunderstanding this? Thanks, Ralph Vince
>>
>> ----- Original Message -----
>> From: "Brian G. Peterson" <[hidden email]>
>> To: "Aleks Clark" <[hidden email]>
>> Cc: <[hidden email]>
>> Sent: Sunday, October 11, 2009 8:46 PM
>> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>>
>>
>> > Aleks Clark wrote:
>> >> As part of a project I'm working on that uses genetic algorithms to
>> >> optimize trading parameters, I find myself seeking a way
>> to evaluate
>> >> the equity curve that results from a given set of trading rules. It
>> >> seems to be an obvious area of research, so I was wondering what's
>> >> available in R-land or just in the world of finance in
>> general. I've
>> >> poked around with splines as a way to express how 'nice' an equity
>> >> curve is (steady upward rise as opposed to a "jagged" line), but I
>> >> feel that there are probably better ways to do things...
>> >
>> > Having worked both in quantitative trading and in more
>> traditional asset
>> > management roles, I've never quite understood the
>> artificial distinction
>> > between "equity curves" and any other kind of returns.  In
>> my experience,
>> > all the usual performance and risk analysis tools (amply
>> provided for in
>> > R) as well as attribution (e.g. Bacon, much of which is
>> implemented in
>> > fPortfolio) are equally applicable to trading strategies as
>> they are to
>> > more traditional investment.  Also see Pat Burns' paper on
>> evaluatinfg
>> > trading strategies for additional ideas.
>> >
>> > Regards,
>> >
>> >   - Brian
>> >
>> > _______________________________________________
>> > [hidden email] mailing list
>> > https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>> > -- Subscriber-posting only.
>> > -- If you want to post, subscribe first.
>>
>> _______________________________________________
>> [hidden email] mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>> -- Subscriber-posting only.
>> -- If you want to post, subscribe first.
>>
>



--
Aleks Clark

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.
Reply | Threaded
Open this post in threaded view
|

Re: Evaluating equity curves

Patrick Burns-2
A traditional measure is the information
ratio.



Patrick Burns
[hidden email]
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of "The R Inferno" and "A Guide for the Unwilling S User")

Aleks Clark wrote:

> Well, the reasoning went like this...since I can instantly look at a
> equity curve and say 'hey that's a nice steady upward slope' as
> opposed to "wow that's a jaggy one with lots of drawdowns", I thought
> to apply some sort of maths to turn this gestalt into a number that
> could be taken along with the final balance as a measure of fitness,
> thus the applying of splines. Now that you mention it however, there
> are probably better ways of evaluating these strategies that aren't
> 'graphical' in nature.
>
> On Mon, Oct 12, 2009 at 6:45 AM, Robert Sams <[hidden email]> wrote:
>> Quite right, Ralph. More precisely: any series of cumulative returns
>> plotted against time on a 2d chart is, *by definition*, an equity curve.
>>
>>
>> Aleks, I'm not sure you're after. To me, the return series of a trading
>> strategy isn't 'evaluated' per se, it's evaluated with respect to some
>> specific questions or conjectures you have regarding it. Articulating
>> this is logically prior to choosing the appropriate mathematical
>> technique.
>>
>> If you can be more specific, we might be able to point you in the right
>> direction.
>>
>> Robert
>>
>>> -----Original Message-----
>>> From: [hidden email]
>>> [mailto:[hidden email]] On Behalf Of R. Vince
>>> Sent: 12 October 2009 02:06
>>> To: Brian G. Peterson; Aleks Clark
>>> Cc: [hidden email]
>>> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>>>
>>> Brian,
>>>
>>> Isn't any stream of cumulative returns, de facto, an equity
>>> curve? Or am I misunderstanding this? Thanks, Ralph Vince
>>>
>>> ----- Original Message -----
>>> From: "Brian G. Peterson" <[hidden email]>
>>> To: "Aleks Clark" <[hidden email]>
>>> Cc: <[hidden email]>
>>> Sent: Sunday, October 11, 2009 8:46 PM
>>> Subject: Re: [R-SIG-Finance] Evaluating equity curves
>>>
>>>
>>>> Aleks Clark wrote:
>>>>> As part of a project I'm working on that uses genetic algorithms to
>>>>> optimize trading parameters, I find myself seeking a way
>>> to evaluate
>>>>> the equity curve that results from a given set of trading rules. It
>>>>> seems to be an obvious area of research, so I was wondering what's
>>>>> available in R-land or just in the world of finance in
>>> general. I've
>>>>> poked around with splines as a way to express how 'nice' an equity
>>>>> curve is (steady upward rise as opposed to a "jagged" line), but I
>>>>> feel that there are probably better ways to do things...
>>>> Having worked both in quantitative trading and in more
>>> traditional asset
>>>> management roles, I've never quite understood the
>>> artificial distinction
>>>> between "equity curves" and any other kind of returns.  In
>>> my experience,
>>>> all the usual performance and risk analysis tools (amply
>>> provided for in
>>>> R) as well as attribution (e.g. Bacon, much of which is
>>> implemented in
>>>> fPortfolio) are equally applicable to trading strategies as
>>> they are to
>>>> more traditional investment.  Also see Pat Burns' paper on
>>> evaluatinfg
>>>> trading strategies for additional ideas.
>>>>
>>>> Regards,
>>>>
>>>>   - Brian
>>>>
>>>> _______________________________________________
>>>> [hidden email] mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>>>> -- Subscriber-posting only.
>>>> -- If you want to post, subscribe first.
>>> _______________________________________________
>>> [hidden email] mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>>> -- Subscriber-posting only.
>>> -- If you want to post, subscribe first.
>>>
>
>
>

_______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only.
-- If you want to post, subscribe first.