HN Theater @HNTheaterMonth

Quantitative finance is a field that takes from many different topics and combines them, so one way to start instead of studying the field directly it helps to learn a handful of different subjects slowly over time.

Most people start with basic programming and basic stock market indicators. Many people start by hacking together some python scripts using some framework and then backtesting to see if their assumptions are correct.

On the more advanced level there is C++, CUDA, and Rust, and similar tech that needs to be fast. When developing a custom algo for an NP problem, you need a lot of speed. On this front, auditing MIT's old Artificial Intelligence class might be a good place to start, which goes over how to optimize NP algorithms and builds up the thought process of how to automate difficult problems. https://www.youtube.com/watch?v=TjZBTDzGeGg&list=PLUl4u3cNGP...

Writing a bot that plays the stock market when boiled down is either 1) pattern matching the market, having a hypothesis, and then writing some code to test that hypothesis or 2) writing a machine that pattern matches for you, finding meta-patterns of sorts. Is closer to Data Science and machine learning.

Quantitative finance can be as simple as what is often done in Data Science work. That is, hacking together some python scripts and testing your assumption, or it can be as complex as being an expert C++ programmer and building up a custom system, combining Software Engineering and Data Science.

Furthermore, in the DS world there is a lot of research at places like https://arxiv.org/ like an existing path of half-solutions that can be applied, tested, and verified. In the Quant world, information is private, so you don't get that luxury. Likewise, most models in the DS world don't work in the Quant world, for at least two reasons: 1) Most machine learning today isn't timeseries based. Eg, a self driving car may frame by frame identify what is in each picture, but it doesn't do a good job retaining data in the larger scene making it difficult to use standard machine learning to solve problems in the financial market. and 2) Most machine learning algos are based on floating point numbers and approximations. When dealing with money you need perfect precision, so even if there is the perfect machine learning library, you might end up writing your own version.

Eh, I'm probably rambling a bit too much here. Either way, quant finance is tons of fun! Also, I've only done projects independently, so I have no idea how different my approaches are to the industry.

Also, another field that might help is DSP, though I admit it's something I don't know enough about, so I can't say with absolute certainty how much it would help.

And of course, there is understanding how the market works itself. This often involves getting your feet wet, usually by paper trading. For many this means learning option trading.

MIT 6.034 Artificial Intelligence, Fall 2010

https://www.youtube.com/watch?v=TjZBTDzGeGg&list=PLUl4u3cNGP...

this one starts with simple stuff MIT 6.034 Artificial Intelligence: https://www.youtube.com/watch?v=TjZBTDzGeGg&list=PLUl4u3cNGP...

you can find all of the videos in this youtube playlist - https://www.youtube.com/watch?v=TjZBTDzGeGg&list=PLUl4u3cNGP...

all lectures are here - https://www.youtube.com/watch?v=TjZBTDzGeGg&list=PLUl4u3cNGP...