One other aside is matlab is much more flexible at vector processing than C or Java. This is very attractive because market data is by nature a discrete time series, and coding various indicators, setup conditions, etc. should be much more straightforward than other OO programming languages. I guess this is what mbatrader is trying to say... RoughTrader
My impression from significant use of both languages is programming in MATLAB is easy for simple things, i.e. data manipulation, or other serial operations which involve heavy mathematics that I can handle with function calls only. I mean, you can use Matlab for complicated things, but then you are likely doing the complicated things with a lot of the built-in functionailty. After doing some more reading, apparently Matlab does have some OO capabilities, so perhaps it can do more than I thought. The lower level languages are still going to be more powerful, but perhaps after learning all the Matlab OO syntax, it wouldn't be as limiting as I first thought. For the other issues mentioned, I would still personally prefer C++, but to each his own. If many languages can do the job required, then it comes down to whatever the programmer(s) is/are most comfortable/experienced with. Memory is definitely a concern, but I think bandwidth (not language-related, obviously) is more of a concern usually. Memory is dirt-cheap relative to bandwidth.
I think it may make such operations easier for the programmer since this functionality is built in and since it is a scripting language, but without checking, I would guess that any and all such vector processing would be available to C and Java programmers in readily available libraries.