Home » Archived » OFMP » OFMP Services
| |
| Re: OFMP Services [message #9063 is a reply to message #9000] |
Wed, 19 March 2008 17:51   |
Eclipse User |
|
|
|
Originally posted by: rgomes1997.yahoo.co.uk
Frederic Conrotte wrote:
> Werner Keil a écrit :
>> Hi,
>>
>> From the creation proposal and my article on OFMP I remember, some
>> Services mentioned there.
>>
>> Do some of those exist already?
>> Especially the Date/Time and Currency Services.
> Yes both exist in current code base, thought they could be extended.
> Date service is about business date management. We have a Calendar
> Service that take a currency calendar as input and
> http://objectlabkit.sourceforge.net/ as underlying engine.
>
> Fred
Hi,
I recap that JQuantLib http://www.jquantlib.org/ is being actively
developed.
We are planning to have european option valuation via Black and Scholes
entering in unit test phase till end of april. At the moment, I'm working
on the core object model which supports all intruments and pricing engines
implemented by QuantLib/C++
In particular, QuantLib(C++) has support for currency calendar and we could
simply implement these classes in JQuantLib(Java) as we already have all
the necessary support classes.
Thanks
--
Richard Gomes
http://wiki.jquantlib.org/index.php/User:RichardGomes
|
|
|
| Re: OFMP Services [message #9083 is a reply to message #9063] |
Thu, 20 March 2008 09:03 |
| Eclipse User |
|
|
|
Originally posted by: frederic.conrotte.kaupthing.lu
Hi Richard
"Richard Gomes" <rgomes1997@yahoo.co.uk> wrote in message
news:frrjpt$um2$1@build.eclipse.org...
> We are planning to have european option valuation via Black and Scholes
> entering in unit test phase till end of april. At the moment, I'm working
> on the core object model which supports all intruments and pricing engines
> implemented by QuantLib/C++
Interesting topic for OFMP.
What source of input do you use to get volatility data ?
What type of volatility do you support ?
Frederic
|
|
|
| Re: OFMP Services [message #9102 is a reply to message #9063] |
Fri, 21 March 2008 15:08 |
 Werner Keil
Messages: 1087
Registered: July 2009 |
Senior Member |
|
|
Richard Gomes wrote:
> Frederic Conrotte wrote:
>
>> Werner Keil a écrit :
>>> Hi,
>>>
>>> From the creation proposal and my article on OFMP I remember, some
>>> Services mentioned there.
>>>
>>> Do some of those exist already?
>>> Especially the Date/Time and Currency Services.
>> Yes both exist in current code base, thought they could be extended.
>> Date service is about business date management. We have a Calendar
>> Service that take a currency calendar as input and
>> http://objectlabkit.sourceforge.net/ as underlying engine.
>>
>> Fred
>
>
> Hi,
>
> I recap that JQuantLib http://www.jquantlib.org/ is being actively
> developed.
>
> We are planning to have european option valuation via Black and Scholes
> entering in unit test phase till end of april. At the moment, I'm working
> on the core object model which supports all intruments and pricing engines
> implemented by QuantLib/C++
>
> In particular, QuantLib(C++) has support for currency calendar and we could
> simply implement these classes in JQuantLib(Java) as we already have all
> the necessary support classes.
>
> Thanks
>
Sounds interesting.
What types of support classes do you have for both the Time/Date aspect
off this and the actual Money/Currency?
I spoke at QCon with several other Spec Leads of the JCP including Rod
Johnson (Spring) One of them leads JSR-310 (Date/Time) and also is the
lead architect of JodaTime (one of the 2 libraries ObjectLabKit already
supports)
So if your code also is compatible with what IS provided by Java itself
on Date/Time or shall be from Java 7 or 8 onwards (basically a
combination of Joda and java.util.Date,... but more in style and design
of Joda's current API) that should be fine.
If you intend to use something homegrown and proprietary, that may not
be favorable for OFMP
And of course, Eclipse has its own sanitary requirements, where accepted
standards like Apache, Spring or JSRs are not a problem, but additional
code often can be. Currency and Money are therefore also better using an
existing JSR proven to work (including Spring, I just did a complete
International Stock Market demo on QCon)
Werner
|
|
| |
| Re: OFMP Services [message #9142 is a reply to message #9124] |
Sun, 30 March 2008 23:42 |
| Eclipse User |
|
|
|
Originally posted by: rgomes1997.yahoo.co.uk
Hi Werner,
Sorry for delay.
I was working a lot on JQuantLib and I had to find some time to answer
properly your email.
Yes... some love Maven, some hate... :)
It's powerful but may be annoying when things go wrong.
About Dates:
JQuantLib is based on QuantLib http://www.quantlib.org.
As the original C++ creature, JQuanLib do not have anything sophisticated as
Joda Time (now JSR 310). Joda supports even leap seconds, which are
additional seconds astronomers decide to insert sometimes in order to keep
our clocks in sync with the God's clock :) ... JQuanLib has only a Date
class which is adequate for dates, I mean, days, months, years.
http://wiki.jquantlib.org/maven2/sites/jquantlib/xref/index. html
JQuantLib is intended to perform price valuation and related tasks which do
not require the accuracy of JSR-310. I dont think I have anything to
contribute regarding JSR-310.
Regarding holidays, QuantLib has very simple methods which respond "yes"
when a certain date is a holiday on a certain country (in fact, on a
certain exchange house). In JQuantLib we will implement pretty much the
same... very simple, very fast.
I have another project which may benefit from JSR-310, as it is intended to
keep historical data and perform a bunch of data mining on it.
Kind Regards
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
Werner Keil wrote:
> Richard,
>
> I just had a brief look at your JavaDocs and XDoc source quotes (using
> Maven is not a bad idea btw, though some people at QCon also criticized
> parts of it ;-)
>
> While I can see nothing on the actual money as such and classes like
> "Quote" are not precise enough (we had plenty of discussions in JSR-275
> about the use of double vs. more precise forms like BigDecimal) a lot of
> those classes show a strong reassemblence of ideas and patterns used by
> this JCP group: JSR 310: Date and Time API.
>
> Therefore I think you might want to get in touch with their Spec Lead
> such as Stephen Colebourne and discuss wheter your ideas may be
> beneficial for this JSR?
>
> Much of it, especially Holiday Calculations may exceed the scope of the
> actual JSR (I'm sure, Stephen knows that) but instead of using some of
> your own proprietary classes "reinventing" Joda or JSR 310 I would
> rather use them (at least in any form of contribution to OFMP)
>
> Werner
|
|
|
| Re: OFMP Services [message #9162 is a reply to message #9102] |
Mon, 31 March 2008 00:06 |
| Eclipse User |
|
|
|
Originally posted by: rgomes1997.yahoo.co.uk
Hi Werner,
Regarding QCon... unfortunately I could not attend.
Next year I will be there!
JQuantLib aims to be a 100% Java 'translation' from QuantLib, which is
written in C++. The work 'translation' does not describe well the amount of
effort necessary for doing this task well. At the moment, it's out of scope
change the original design and concepts adopted by QuantLib.
There are other things to consider. JQuantLib aims to be extremely fast and
strong type-checked (at the same time!) which basically, means we are using
primitive types wherever possible and also using JSR-308 in order to obtain
maximum semantic type checking at compile time.
http://www.jquantlib.org/index.php/DesignTypeChecking
http://www.jquantlib.org/index.php/DesignPerformance
http://www.jquantlib.org/index.php/Providing_unmutability_to _receivers
In a nutshell, we try to avoid object creation/destruction as possible,
replace Collections by arrays of primitive types and related techniques.
http://fastutil.dsi.unimi.it/docs/
Regarding dates, etc... we are not able to adopt JSR-310 at the moment.
In the future, we can use different implementations of a certain interface
as necessary, aiming performance or accuracy, as necessary.
I would like to comply with requirements Eclipse and Apache projects but I
dont think it's a feasible reality at the moment because we are a very
small team with huge challenges. You can see our short-term milestones at:
http://www.jquantlib.org/index.php/JQuantLib:Community_Porta l
Thanks
-- Richard
Werner Keil wrote:
> Richard Gomes wrote:
>> Frederic Conrotte wrote:
>>
>>> Werner Keil a écrit :
>>>> Hi,
>>>>
>>>> From the creation proposal and my article on OFMP I remember, some
>>>> Services mentioned there.
>>>>
>>>> Do some of those exist already?
>>>> Especially the Date/Time and Currency Services.
>>> Yes both exist in current code base, thought they could be extended.
>>> Date service is about business date management. We have a Calendar
>>> Service that take a currency calendar as input and
>>> http://objectlabkit.sourceforge.net/ as underlying engine.
>>>
>>> Fred
>>
>>
>> Hi,
>>
>> I recap that JQuantLib http://www.jquantlib.org/ is being actively
>> developed.
>>
>> We are planning to have european option valuation via Black and Scholes
>> entering in unit test phase till end of april. At the moment, I'm working
>> on the core object
modehttp://www.jquantlib.org/index.php/JQuantLib:Community_P ortall which
supports all intruments and pricing
>> engines implemented by QuantLib/C++
>>
>> In particular, QuantLib(C++) has support for currency calendar and we
>> could simply implement these classes in JQuantLib(Java) as we already
>> have all the necessary support classes.
>>
>> Thanks
>>
>
> Sounds interesting.
> What types of support classes do you have for both the Time/Date aspect
> off this and the actual Money/Currency?
>
> I spoke at QCon with several other Spec Leads of the JCP including Rod
> Johnson (Spring) One of them leads JSR-310 (Date/Time) and also is the
> lead architect of JodaTime (one of the 2 libraries ObjectLabKit already
> supports)
>
> So if your code also is compatible with what IS provided by Java itself
> on Date/Time or shall be from Java 7 or 8 onwards (basically a
> combination of Joda and java.util.Date,... but more in style and design
> of Joda's current API) that should be fine.
> If you intend to use something homegrown and proprietary, that may not
> be favorable for OFMP
>
> And of course, Eclipse has its own sanitary requirements, where accepted
> standards like Apache, Spring or JSRs are not a problem, but additional
> code often can be. Currency and Money are therefore also better using an
> existing JSR proven to work (including Spring, I just did a complete
> International Stock Market demo on QCon)
>
> Werner
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
|
|
|
| Re: OFMP Services [message #9182 is a reply to message #9083] |
Mon, 31 March 2008 00:30 |
| Eclipse User |
|
|
|
Originally posted by: rgomes1997.yahoo.co.uk
Hi Frederic,
Sorry for the long delay.
I was working hard on JQuantLib and I had to find enough time to answer your
question properly.
In fact JQuantLib is simply a price valuation engine (among other things!)
and it is not meant to obtain/keep/manage historical data.
You can obtain volatility from historical data and pass this data to
JQuantLib in order to obtain the 'fair' price of an option, for instance.
In adition, JQuantLib also supports implied volatility, which is 'what the
market behaves the actual volatility is'. This value is based on
* strike price
* maturity (expiry date)
* current price of underlying instrument
* 'zero-risk' interest rate.
Feel free to contact me directly.
Kind Regards
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
Frederic Conrotte wrote:
> Hi Richard
>
> "Richard Gomes" <rgomes1997@yahoo.co.uk> wrote in message
> news:frrjpt$um2$1@build.eclipse.org...
>> We are planning to have european option valuation via Black and Scholes
>> entering in unit test phase till end of april. At the moment, I'm working
>> on the core object model which supports all intruments and pricing
>> engines implemented by QuantLib/C++
>
> Interesting topic for OFMP.
>
> What source of input do you use to get volatility data ?
> What type of volatility do you support ?
>
> Frederic
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
|
|
| | | |
| Re: OFMP Services [message #568393 is a reply to message #9063] |
Fri, 21 March 2008 15:08 |
Werner Keil
Messages: 1087
Registered: July 2009 |
Senior Member |
|
|
Richard Gomes wrote:
> Frederic Conrotte wrote:
>
>> Werner Keil a écrit :
>>> Hi,
>>>
>>> From the creation proposal and my article on OFMP I remember, some
>>> Services mentioned there.
>>>
>>> Do some of those exist already?
>>> Especially the Date/Time and Currency Services.
>> Yes both exist in current code base, thought they could be extended.
>> Date service is about business date management. We have a Calendar
>> Service that take a currency calendar as input and
>> http://objectlabkit.sourceforge.net/ as underlying engine.
>>
>> Fred
>
>
> Hi,
>
> I recap that JQuantLib http://www.jquantlib.org/ is being actively
> developed.
>
> We are planning to have european option valuation via Black and Scholes
> entering in unit test phase till end of april. At the moment, I'm working
> on the core object model which supports all intruments and pricing engines
> implemented by QuantLib/C++
>
> In particular, QuantLib(C++) has support for currency calendar and we could
> simply implement these classes in JQuantLib(Java) as we already have all
> the necessary support classes.
>
> Thanks
>
Sounds interesting.
What types of support classes do you have for both the Time/Date aspect
off this and the actual Money/Currency?
I spoke at QCon with several other Spec Leads of the JCP including Rod
Johnson (Spring) One of them leads JSR-310 (Date/Time) and also is the
lead architect of JodaTime (one of the 2 libraries ObjectLabKit already
supports)
So if your code also is compatible with what IS provided by Java itself
on Date/Time or shall be from Java 7 or 8 onwards (basically a
combination of Joda and java.util.Date,... but more in style and design
of Joda's current API) that should be fine.
If you intend to use something homegrown and proprietary, that may not
be favorable for OFMP
And of course, Eclipse has its own sanitary requirements, where accepted
standards like Apache, Spring or JSRs are not a problem, but additional
code often can be. Currency and Money are therefore also better using an
existing JSR proven to work (including Spring, I just did a complete
International Stock Market demo on QCon)
Werner
|
|
| |
| Re: OFMP Services [message #568421 is a reply to message #9124] |
Sun, 30 March 2008 23:42 |
Richard Gomes
Messages: 15
Registered: June 2010 |
Junior Member |
|
|
Hi Werner,
Sorry for delay.
I was working a lot on JQuantLib and I had to find some time to answer
properly your email.
Yes... some love Maven, some hate... :)
It's powerful but may be annoying when things go wrong.
About Dates:
JQuantLib is based on QuantLib http://www.quantlib.org
As the original C++ creature, JQuanLib do not have anything sophisticated as
Joda Time (now JSR 310). Joda supports even leap seconds, which are
additional seconds astronomers decide to insert sometimes in order to keep
our clocks in sync with the God's clock :) ... JQuanLib has only a Date
class which is adequate for dates, I mean, days, months, years.
http://wiki.jquantlib.org/maven2/sites/jquantlib/xref/index. html
JQuantLib is intended to perform price valuation and related tasks which do
not require the accuracy of JSR-310. I dont think I have anything to
contribute regarding JSR-310.
Regarding holidays, QuantLib has very simple methods which respond "yes"
when a certain date is a holiday on a certain country (in fact, on a
certain exchange house). In JQuantLib we will implement pretty much the
same... very simple, very fast.
I have another project which may benefit from JSR-310, as it is intended to
keep historical data and perform a bunch of data mining on it.
Kind Regards
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
Werner Keil wrote:
> Richard,
>
> I just had a brief look at your JavaDocs and XDoc source quotes (using
> Maven is not a bad idea btw, though some people at QCon also criticized
> parts of it ;-)
>
> While I can see nothing on the actual money as such and classes like
> "Quote" are not precise enough (we had plenty of discussions in JSR-275
> about the use of double vs. more precise forms like BigDecimal) a lot of
> those classes show a strong reassemblence of ideas and patterns used by
> this JCP group: JSR 310: Date and Time API.
>
> Therefore I think you might want to get in touch with their Spec Lead
> such as Stephen Colebourne and discuss wheter your ideas may be
> beneficial for this JSR?
>
> Much of it, especially Holiday Calculations may exceed the scope of the
> actual JSR (I'm sure, Stephen knows that) but instead of using some of
> your own proprietary classes "reinventing" Joda or JSR 310 I would
> rather use them (at least in any form of contribution to OFMP)
>
> Werner
|
|
|
| Re: OFMP Services [message #568435 is a reply to message #9102] |
Mon, 31 March 2008 00:06 |
Richard Gomes
Messages: 15
Registered: June 2010 |
Junior Member |
|
|
Hi Werner,
Regarding QCon... unfortunately I could not attend.
Next year I will be there!
JQuantLib aims to be a 100% Java 'translation' from QuantLib, which is
written in C++. The work 'translation' does not describe well the amount of
effort necessary for doing this task well. At the moment, it's out of scope
change the original design and concepts adopted by QuantLib.
There are other things to consider. JQuantLib aims to be extremely fast and
strong type-checked (at the same time!) which basically, means we are using
primitive types wherever possible and also using JSR-308 in order to obtain
maximum semantic type checking at compile time.
http://www.jquantlib.org/index.php/DesignTypeChecking
http://www.jquantlib.org/index.php/DesignPerformance
http://www.jquantlib.org/index.php/Providing_unmutability_to _receivers
In a nutshell, we try to avoid object creation/destruction as possible,
replace Collections by arrays of primitive types and related techniques.
http://fastutil.dsi.unimi.it/docs/
Regarding dates, etc... we are not able to adopt JSR-310 at the moment.
In the future, we can use different implementations of a certain interface
as necessary, aiming performance or accuracy, as necessary.
I would like to comply with requirements Eclipse and Apache projects but I
dont think it's a feasible reality at the moment because we are a very
small team with huge challenges. You can see our short-term milestones at:
http://www.jquantlib.org/index.php/JQuantLib:Community_Porta l
Thanks
-- Richard
Werner Keil wrote:
> Richard Gomes wrote:
>> Frederic Conrotte wrote:
>>
>>> Werner Keil a écrit :
>>>> Hi,
>>>>
>>>> From the creation proposal and my article on OFMP I remember, some
>>>> Services mentioned there.
>>>>
>>>> Do some of those exist already?
>>>> Especially the Date/Time and Currency Services.
>>> Yes both exist in current code base, thought they could be extended.
>>> Date service is about business date management. We have a Calendar
>>> Service that take a currency calendar as input and
>>> http://objectlabkit.sourceforge.net/ as underlying engine.
>>>
>>> Fred
>>
>>
>> Hi,
>>
>> I recap that JQuantLib http://www.jquantlib.org/ is being actively
>> developed.
>>
>> We are planning to have european option valuation via Black and Scholes
>> entering in unit test phase till end of april. At the moment, I'm working
>> on the core object
modehttp://www.jquantlib.org/index.php/JQuantLib:Community_P ortall which
supports all intruments and pricing
>> engines implemented by QuantLib/C++
>>
>> In particular, QuantLib(C++) has support for currency calendar and we
>> could simply implement these classes in JQuantLib(Java) as we already
>> have all the necessary support classes.
>>
>> Thanks
>>
>
> Sounds interesting.
> What types of support classes do you have for both the Time/Date aspect
> off this and the actual Money/Currency?
>
> I spoke at QCon with several other Spec Leads of the JCP including Rod
> Johnson (Spring) One of them leads JSR-310 (Date/Time) and also is the
> lead architect of JodaTime (one of the 2 libraries ObjectLabKit already
> supports)
>
> So if your code also is compatible with what IS provided by Java itself
> on Date/Time or shall be from Java 7 or 8 onwards (basically a
> combination of Joda and java.util.Date,... but more in style and design
> of Joda's current API) that should be fine.
> If you intend to use something homegrown and proprietary, that may not
> be favorable for OFMP
>
> And of course, Eclipse has its own sanitary requirements, where accepted
> standards like Apache, Spring or JSRs are not a problem, but additional
> code often can be. Currency and Money are therefore also better using an
> existing JSR proven to work (including Spring, I just did a complete
> International Stock Market demo on QCon)
>
> Werner
--
Richard Gomes
http://wiki.jquantlib.org/index.php/RichardGomes
|
|
| |
Goto Forum:
Current Time: Thu Apr 25 02:17:30 GMT 2024
Powered by FUDForum. Page generated in 0.04695 seconds |
] 
Search
Help
Register
Login
Home
