Has anyone used Python with real-time data, such as 5 second bars, NOT with Interactive Brokers, Bloomberg or Quantopian? Have most people here used end-of-day data with Python?
Are Python compilers generally optimizing worser to assembly and running worse than C/C++ (mingw , gcc , visual studio x64 etc) or it depend also on ways of coding? I have 2-3 years of programming experience and have gotten best multithreaded C++ performance on linux, gcc compiler. On windows compilers all appear to perform almost equal to eachother with single thread.
The lecturer forgot to mention another reason why the CIS dept at Stanford has more honor code and plagiarism violations than all other departments combined is to a large degree Asian students. Of course he cannot say that as political correctness is still the forte at US colleges. But truth be told students who come from HK or China and even Japan or Korea in majority have never heard the term honor code or thought of copying as cheating. Most can be blamed because they are intelligent enough to know, a few maybe can't be faulted. But fact is in China and other Asian countries as long as you get your A it does not matter how you came to it. Sorry, unrelated to Python but as the posted lecture did not focus on Python but honor code I thought I wanted to inject this point. It equally pertains to trading and finance in general. There is so much blatant cheating going on in stock markets around the world and nobody really is competent enough out there to properly police it.
There are several Python graphics packages. Have you tried PyQtGraph? It is mentioned in "20 Python Libraries You Aren't Using (But Should)." It is also mentioned this podcast "Episode #77: 20 Python Libraries You Aren't Using (But Should)." It can be added in the Anaconda Python install.
I was wondering how many people are using real-time data outside of the standard broker client programs. I am under the impression most people access end-of-day data from free sources such as Yahoo or Google. What is transformed exchange data? How much does it cost? If you don't mind me asking.
packages like this sound nice and are pleasing to the eye but when you actually get to the gist of it they are not applicable to a financial or trading application, at least not one that requires dynamic charting. If you want to know what I mean then take a look at this and how any Java, any C++, any Python charting library measures up to it: https://www.scichart.com/wpf-chart-examples/ Maybe not all applications require dynamic graphs or charting but when you do then I have never seen a library that comes even close to the C#/WPF library I linked to above. There is a reason more and more front-end applications in banks and hedge funds are coded in C# and use WPF/xaml for visuals.
Raw ticks. I can tell you more via PM. Python is very good for it. So, petty much any serious quant shop runs their R&D on python these days. There is really nothing like pandas out there. Matlab and R are close second, but the sheer horsepower that pandas+numpy deliver are unmatched. Plenty of people run live execution straight from Python too if they are using broker algos and they don't care for latency much. At banks traders rarely program much, use BBG charts and have excel-based tools that they use for their analytics. It's ghetto, but it gives you flexibility and fast deployment time. C++ tools are usually risk viewers and all sort of execution engines. This brings me to my main point. Traders job is to extract alpha, not write pretty widgets or charts. Unless you are sensitive to latency or have other performance issues, Python is more then enough.
I was not trying to suggest traders would be writing applications. IT departments at banks do, however. And I would claim every bank in the top 10 league tables provides traders with some applications that were written in-house either in Java or C#, or both. I am talking about front-end applications. And I only mention it because Python's charting libraries were brought up. Python lags badly behind in terms of visualization. Pending need and requirements this might be a non-issue or a deal breaker.