Empirical is statically typed. I can run a backtest overnight knowing it won't crash because of a type error or misspelled column name.

I'm currently in the (long) process of adding streaming computation. Once that's ready, Empirical will be able to handle everything from gargantuan CSV files to real-time data using the same language primitives.

I talked about some of these goals when I launched on HN last year:

https://news.ycombinator.com/item?id=19969390