I've been rewriting most of my financial analytics code in Scala & its a genuine pleasure to see code sizes reduce 5-6 times...
What did you write your analytics code in before?! Most quants use R, Matlab, q/kdb+, OCaml, Python, etc. I can't imagine an analyst would ever use a language like Java or C++ for research.
M$ Excel, M$ SQL Server, M$ VBA macros
Also known as: what most of Wall Street uses. Are you sure you want a career in this industry?
what about front-end solutions for this newfangled language, does Scala do reporting, does Scala talk to Bloomberg, do other IB's use Scala
All of which are legitimate concerns. Did you bother to address any of those?
As the others have mentioned, banks aren't looking for "fast" or even "productive"; they're looking for "reliable". Wall Street doesn't chase technology the same way Silicon Valley does. Remember the FourSquare outage (MongoDB) or the numerous issues with Twitter back in the day (Ruby on Rails)? Banks need to have better uptime than that.
We used to talk longingly about using strongly typed languages when I was at an IB. We mostly used perl for munging data and SAS for actual heavy lifting. It would have been great to have the compiler yell at me for adding a return to a dollar value or whatever.
Were you at ETL? The prop group I was in at Morgan used q/kdb+, which was great for a lot of things but did cause some trip-ups that static typing would have prevented. My group eventually shut down and I left the firm, but I had heard Morgan was looking into F# to use company-wide. (I used Ada in a previous life, so I know full-well the benefits of strong typing.)
Dude, chill. I have worked on Wall St for over a decade at this point . There is a huge, HUGE diversity. Some use toolsets they are comfortable with, others with what they get interested in ( after it crosses a certain stability threshold ofcourse ). BTW R itself gained mainstream acceptance in 2009+. I can imagine you saying in 2008 no quant in their right minds would use R, its just too new. It was new then, but it's defacto now. That was true of OCaml which Jane St uses almost everywhere now but didn't do so just a few years back. Ofcourse there are quants using C++ for all their risk analytics - 50% of my classmates got hired on that very credential! As to Java, I have personally written trading software for IBs in Java way back in 2002 that used pricing modules for all the instruments also written in Java. Guess how mainstream Java was back then ?
Finally, there are actually fairly decent risk analytics DSLs in Scala out there.
Other than reporting ( you really can't compete with SAP/Crystal Reports on that front), solutions in Scala address the rest of your legitimate concerns using DCOM hooks to pre-existing C libraries & DLLs. Why reinvent the wheel ?
I don't see why you are getting so worked up about my "inexperienced rant". I simply described a real-life situation I encountered which I felt was funny viz. highly risk-taking alpha personalities getting hugely risk-averse when it comes to something quite innocent like language adoption.
The reason I criticized your post (and I'm not the only one)
is that I don't want a young'un to read this and believe that's the right way to be thinking about career or business decisions.
You interviewed somewhere that isn't for you. So try someplace else. Whining on the Internet about "teh M$" doesn't make you look like a brazen intellectual.
Argh! If I'm in the midwest now can't I have worked in NYC before I moved here (which is what happened btw, now that you are so curious about my antecedents). And M has taken so many $ from me in the form of numerous OS updates & backups that I can justifiably call them M$ :) Should one be an intellectual to play around with scala ? Jeez.
Alright, truce. I thought I had remembered a Scala career question on Quant Stack Exchange (I'm a moderator there). So there are a couple finance shops that use it:
Comments
Your rant screams inexperience.
I've been rewriting most of my financial analytics code in Scala & its a genuine pleasure to see code sizes reduce 5-6 times...
What did you write your analytics code in before?! Most quants use R, Matlab, q/kdb+, OCaml, Python, etc. I can't imagine an analyst would ever use a language like Java or C++ for research.
M$ Excel, M$ SQL Server, M$ VBA macros
Also known as: what most of Wall Street uses. Are you sure you want a career in this industry?
what about front-end solutions for this newfangled language, does Scala do reporting, does Scala talk to Bloomberg, do other IB's use Scala
All of which are legitimate concerns. Did you bother to address any of those?
As the others have mentioned, banks aren't looking for "fast" or even "productive"; they're looking for "reliable". Wall Street doesn't chase technology the same way Silicon Valley does. Remember the FourSquare outage (MongoDB) or the numerous issues with Twitter back in the day (Ruby on Rails)? Banks need to have better uptime than that.
> Your rant screams inexperience.
So does the "M$" spelling...
We used to talk longingly about using strongly typed languages when I was at an IB. We mostly used perl for munging data and SAS for actual heavy lifting. It would have been great to have the compiler yell at me for adding a return to a dollar value or whatever.
Were you at ETL? The prop group I was in at Morgan used q/kdb+, which was great for a lot of things but did cause some trip-ups that static typing would have prevented. My group eventually shut down and I left the firm, but I had heard Morgan was looking into F# to use company-wide. (I used Ada in a previous life, so I know full-well the benefits of strong typing.)
Yup. Heard they were doing the F# thing too, but can't see them getting away from unix...
Dude, chill. I have worked on Wall St for over a decade at this point . There is a huge, HUGE diversity. Some use toolsets they are comfortable with, others with what they get interested in ( after it crosses a certain stability threshold ofcourse ). BTW R itself gained mainstream acceptance in 2009+. I can imagine you saying in 2008 no quant in their right minds would use R, its just too new. It was new then, but it's defacto now. That was true of OCaml which Jane St uses almost everywhere now but didn't do so just a few years back. Ofcourse there are quants using C++ for all their risk analytics - 50% of my classmates got hired on that very credential! As to Java, I have personally written trading software for IBs in Java way back in 2002 that used pricing modules for all the instruments also written in Java. Guess how mainstream Java was back then ? Finally, there are actually fairly decent risk analytics DSLs in Scala out there.
Other than reporting ( you really can't compete with SAP/Crystal Reports on that front), solutions in Scala address the rest of your legitimate concerns using DCOM hooks to pre-existing C libraries & DLLs. Why reinvent the wheel ?
I don't see why you are getting so worked up about my "inexperienced rant". I simply described a real-life situation I encountered which I felt was funny viz. highly risk-taking alpha personalities getting hugely risk-averse when it comes to something quite innocent like language adoption.
I have worked on Wall St for over a decade at this point.
And yet you say in this post that you're in the midwest:
http://news.ycombinator.com/item?id=2507002
The reason I criticized your post (and I'm not the only one) is that I don't want a young'un to read this and believe that's the right way to be thinking about career or business decisions.
You interviewed somewhere that isn't for you. So try someplace else. Whining on the Internet about "teh M$" doesn't make you look like a brazen intellectual.
Argh! If I'm in the midwest now can't I have worked in NYC before I moved here (which is what happened btw, now that you are so curious about my antecedents). And M has taken so many $ from me in the form of numerous OS updates & backups that I can justifiably call them M$ :) Should one be an intellectual to play around with scala ? Jeez.
Alright, truce. I thought I had remembered a Scala career question on Quant Stack Exchange (I'm a moderator there). So there are a couple finance shops that use it:
http://quant.stackexchange.com/q/298/35