Browsing back over the archives I'm struck by the numbers of views on various topics range from a handful to almost 100,000. The higher ones tend to be things like VBA/C++ problems. Excel plugins and the like. How to do bits and pieces in matlab or Splus/R. How to do garch in Java. C++ problems in general.Got me wondering. Where is the pain for quants when it comes to software development? Where are the integration points? Are people writing lots of excel plugins? Server side C++ with Java or VB frontends on client? Can someone shed some light? Generalizations fine - its friday morning afterall.

QuoteOriginally posted by: stuntprogrammerGot me wondering. Where is the pain for quants when it comes to software development? Where are the integration points? Are people writing lots of excel plugins? Server side C++ with Java or VB frontends on client? Can someone shed some light? Generalizations fine - its friday morning afterall.I tried a number of threads on these topics but the response is luke-warm, UML, patterns, maintainability stuff.Many of the posts are related to problems that need to be solved now, which is fair enough. Writing code is easy; it's what happens after it has been written is when the fun starts. Is that what you are referring to?(patch22 has some good posts relating to the s/w side of QF app integration.)

QuoteOriginally posted by: CuchulainnWriting code is easy; it's what happens after it has been written is when the fun starts. Is that what you are referring to?Yes; broad overview of what type of code is being written and what its being integrated into by the quants.ie. what % of quants write c++ models or functions that they then call from excel? what are the other %s doing?

For me the pain is always re-inventing the wheel - it's so tempting to look back at the various platforms that I have live, and think - "re-factor or re-hash" rather then "re-use".My current favoured solutions tend to be almost like macro programming, using python, but delegating to more powerful tools such as R for stats and django (web framework) for presentation.

why are we bothering with this coding?it's boring, superfluous and should be outsourced.we are wasting our time.

QuoteOriginally posted by: chromowhy are we bothering with this coding?it's boring, superfluous and should be outsourced.we are wasting our time.Right, when can I start please? It is an inalienable right not to want to program if you don't want but outsourcing is less than optimal because too many people who know just pieces of the puzzle. Chain iof command problems ensue.For the technical algorithms: the formulation, design and coding should be done by one person who knows QF and CS.

QuoteOriginally posted by: stuntprogrammerQuoteOriginally posted by: CuchulainnWriting code is easy; it's what happens after it has been written is when the fun starts. Is that what you are referring to?Yes; broad overview of what type of code is being written and what its being integrated into by the quants.ie. what % of quants write c++ models or functions that they then call from excel? what are the other %s doing?A % is difficult but I would guess that it is in the range [40,70] of quants use C++ in one form or another. And they are all interested in C++-Excel interop. Maybe Dominic could give a more accurate estimate.I get the impression that VBA is used a lot.C# is being used here and there.(this is not including the middleware s/w)BTW how popular is Fortran, or would we rather not talk about it

Could I suggest the question "how to resolve the communication/information/synchornisation problem"in place of the "how to code better problem".CS and QF are too broad for one mind to make a reasonable job of, and the psychology required to master those is not going to be very interested in 'operations and support'.Need to draw some lines. Put the correct people in between the appropriate lines and enable the communication across said lines. [motherhood statement]I'm drawing my first line in a controversial place. Beyond programming.Why not?

QuoteOriginally posted by: chromoCould I suggest the question "how to resolve the communication/information/synchornisation problem"in place of the "how to code better problem".CS and QF are too broad for one mind to make a reasonable job of, and the psychology required to master those is not going to be very interested in 'operations and support'.Need to draw some lines. Put the correct people in between the appropriate lines and enable the communication across said lines. [motherhood statement]I'm drawing my first line in a controversial place. Beyond programming.Why not?I agree. If we could see more of these discussions.If I could say it like: how to improve the workflow from model to product, that kind of idea? Reduce impedance mismatch? One does not have to know all QF/IT/maths, just the interfaces with the other stakeholders.

Mr Stunt has some of the answer: scripting. Just need something a little stronger than scripting (or is python enough?). Some chap was calling VBA a "scripting excuse hidden behind a grid control" ('miserable' and expletives omitted).I could take the fifth "it is the responsiblity of management to allocate the tasks appropriately and efficiently" so it's not our problem we get tasked with so much coding. We just do what is asked for. Qs out there have got used to being asked for certain classes of solution and have got better at making those. The hands on part of management has not historically been a strong point of financial institutions. The part of managment that involves owning a yatch has. - Fire the leaders, fire the cannons, fire in the hold.