Also, C++ today is equivalent to template metaprogramming black magic. If you can't do it at the level of maintaining boost MPL, you're regarded as an utter failure who's better playing with some toy language instead of calling himself a programmer. Read Zed Shaw's rant on C++: http://librelist.com/browser//mongrel2/2010/7/15/c-verses-c++/#770d94bcfc6ddf1d8510199996b607dd "Remember the Plague of 2000? If I wanted to fry my brain trying to figure out how to add two numbers with templates I'd go use LISP".
Nice piece from my former fund on why they chose to use python: https://www.ahl.com/ahl-tech-why-python Only a slight inaccuracy (or more kindly, a simplification): "Previously, research on models would be done in R, and the model would be rewritten in Java for production." Actually it was worse than that: research and production was actually a pot pourri of R, Java, Matlab, S-plus, C++ and an in-house proprietary language. GAT
I find using C# and calling R for the data science that is hard to do in C#, to be a very nice combination.
If speed is important then you are better off with C, which is much faster than C++. I don't see the point of C++. Most applications are overspected. With better design speed wouldn't be much of a problem.