MATLAB for quantitative finance

Current License: CC BY-SA 4.0

12 events

when toggle format	what		by	license	comment
May 19, 2020 at 14:01	comment	added	Stéphane		@jbs Eg.: To price options, I might have to use Monte Carlo simulations. It starts to matter when you have tens of thousands of contracts to price inside a maximum likelihood estimation. You can vectorize some of the work to speed it up, but you cannot vectorize everything in Python -- you'll have to work with loops somewhere for this. Alternatively, you can call compiled C++ code to do it and, chances are, that code will run WAY faster. That's why people use Numba in Python or Mex files in MATLAB.
May 19, 2020 at 11:01	comment	added	user24980		I though Numpy was already written in C, don't understand the point of Numba?
May 19, 2020 at 5:42	answer	added	chrisaycock		timeline score: 0
Apr 24, 2020 at 18:47	vote	accept	Stéphane
Apr 22, 2020 at 11:44	comment	added	amdopt		Using the Numba package (numba.pydata.org) to compile mathematical functions in C would be the way to go. You get the speed of C within Python. Also, FWIW, Python isn't just for scripting. If you really want to build all the objects from scratch, OOP in Python is no different in principle than in C++.
Apr 22, 2020 at 9:00	history	tweeted			twitter.com/StackQuant/status/1252884884298715137
Apr 22, 2020 at 8:06	comment	added	Lisa Ann		C++ is the fastest when it comes to Monte Carlo simulations, and you can easily compile and source your functions in R through the excellent Rcpp package. The other way is to use Python and JIT compilation through Numba, which should be just a bit slower than C++ but doesn't require to switch between two different languages. OOP is nice here, but if you use QuantLib-Python you already have very structured objects: you can relax and enjoy some basic functional programming design.
Apr 21, 2020 at 22:21	answer	added	ir7		timeline score: 1
Apr 21, 2020 at 20:36	history	edited	Stéphane	CC BY-SA 4.0	deleted 2 characters in body
Apr 21, 2020 at 20:16	answer	added	Oscar		timeline score: 2
Apr 21, 2020 at 20:07	comment	added	Stéphane		Most of what I do involves econometrics. I did some work in macroeconomic forecasting using machine learning. It was uber time consumming -- we're talking about half a million forecasting errors in the main backtest across models, variables and horizons. All of my current projects involve econometrics and most of my work will involve a heavy empirical element or simulations... So, econometrics, mostly.
Apr 21, 2020 at 19:15	history	asked	Stéphane	CC BY-SA 4.0