I'm interested in exploring algorithm programming for use in proprietary trading. I trade in conventional ways and have a few clients, but no formal training in quantitative finance or C++ programming. I do know a bit about finance and have a fairly good math background.I applied to Cass Quant Finance MsC a few years ago, didn't get accepted,and I really wouldn't like to try it again. My intention is not to work at a bank or hedge fund, but to develop a winning tool for the markets. I was wondering if it's possible to learn it on my own through books or small courses, or if it's only possible in a job or going to a full university program. Please let me know your opinions and I'd really like some books and courses recommendations. I heard that QuantNet is good for C++ and I know a few books, but not much else.regards

University is neither necessary not sufficient in order to be a good progammer. Many of the best programmers learned but just doing it at home before/during entering industry. A good maths background can be useful, but not a guarantee of being a good programmer.I developed the Quantnet C++ course and you get very good support from the teaching assistants. // Now that you ask about courses, I feel it incumbent on me to say that there is a keenly priced C++ course for students coming up next month at Birmingham University

I'm really not familiar with the area, and I'm not sure if C++ is required, but from what I heard in the forums, everyone seems to know a bit...Maybe join something like Quantopian, read some essential books and slowly progress?I'd love to take the course in England, but I live very far from there and it would be pretty innefective cost wise for me going for just a 3 day course, but thanks for the recommendation.

I would advise also to look at Matlab.

A young person's guide to the algorithms

Quote* If you want to learn about how that works (instead if how to use it), then "Numetical Recipes in (any language, probably C) is a good starts: it covers a lot of numerical computing themes.To actually learn numerical analysis the following are 60's/70's classicsDe Boor and ConteDahlquist and BjorkHerbert KellerIsaacson and KellerHilderbrandHenrici Stoer and BulirschGolub and van LoanHageman and YoungMarchukGodunovand ..

QuoteOriginally posted by: outrunYes I agree.I'll try to make a venn diagram that shows the 7 different aspects of learning + {numerical, algorithm, programming} Good idea! So, from A - Z? Specially, how to explain to someone who has never programmed.One of the issues I see it that novices start programming before they have formulated the alogorithm in their head, e.g. sum the values in a map<string, int>A 1.0z 3.03C -1.873etc.They spend days debugging because they get confucius between strings, ints, iterators and maps.

QuoteI'm interested in exploring algorithm programming for use in proprietary trading. I trade in conventional ways and have a few clients, but no formal training in quantitative finance or C++ programming. I do know a bit about finance and have a fairly good math background.Algo programming might be a small bridge too far just now. Learn the fundamentals of C++.

Quote make a software design, ..this is the bit that Cuch likes As do many other designers For small programs you can get away with just coding. But for 10-man 10-year s/w projects you need something more. 3 levels1. Get it working (aka hacking)2. Get it right3. Get it optimisedThe steps 1-->3 is very common (i.e. skip step 2). That's why we have the software crisis and so many code maintainers.End of sermon, Daniel XII // do you know this book (good one)?

QuoteHere you'll again have choices: both design choices (should I use certain language features or not -like C++ templates-) These are language choices, not design choices. Templates are irrelevant during during analysis and design. Mind you, these are useful during implementation.

Quote you need to builds a CDS model in Matlab. You don't need to be a good coder, nor worry about numerical stuff. The focus is making a model that your Quant manager like (and which is of course always wrong if you look at the real word)Sounds like "I don't need to know nothing?" It means moving to a predominantly (professional) software team will be almost impossible.Quote* you need to build a high performance algo trading bot. You need to be a very good coder, know a lot about hardware architecture and have a lot of knowledge about various APIs and 3rd party libraries. > 10 years experience?I think OP wants the second scenario.