What kind of programming will I need to know for quantitative trading?

[SixteenTick]

For context, I am going to a quantitative trading internship soon, but am not sure what kind of coding I will be required to do. Could anyone shed more light on what the day-to-day looks like for a trader regarding what they have to code?

 

[Daniel Duffy]

My guess

C++/C
Python
Java

VBA/Excel??

 

[Paul Lopez]

Learn how to implement as clean and as efficient code as you can. That'll serve you best for any endeavor, especially quant or HF/low latency focused work.
Enjoy the internship!

 

[Paul Lopez]

Solid math and comp sci basics never get old as I'm still learning myself!

 

[MichaelLewis]

Perhaps a bit beyond the scope for right now, but things they don’t really teach in school that are really important if you ask me.

Network optimization
Kernal bypass

Edit:
I’d also add that you’ll want a good understanding of C++ concepts and state machines.

 

[CrossGamma]

For context, I am going to a quantitative trading internship soon, but am not sure what kind of coding I will be required to do. Could anyone shed more light on what the day-to-day looks like for a trader regarding what they have to code?

Python, Python and probably some more Python.

 

[Daniel Duffy]

Python, Python and probably some more Python.

What do you do for performance?

 

[CrossGamma]

What do you do for performance?

At the bigger shops, quant traders mostly work on analysis and very rarely on performance-sensitive code. The latency stack is mostly owned by developers.

 

[Daniel Duffy]

Of course, Python doesn't engender good programming practices. Eventually, it becomes unmaintainable which may or may not be serious. But for POC and throwaway prototypes it is a good scripting language.

 

[Carlos Cabrera Diaz]

I was always confused why in Python you have to import a library to create abstract classes (and what was it anyways?) before I understood pure virtual functions

 

[Daniel Duffy]

I was always confused why in Python you have to import a library to create abstract classes (and what was it anyways?) before I understood pure virtual functions

In my experience 90% of OOP in Python is a load of (unmaintainable) junk. I can give at least 10 reasons.

On the other hand, Python OOP is a nice medium when you want to learn how design patterns work. Having done that, you can port/rewrite the patterns in C++.

 

[Paul Lopez]

In my experience 90% of OOP in Python is a load of (unmaintainable) junk. I can give at least 10 reasons.

On the other hand, Python OOP is a nice medium when you want to learn how design patterns work. Having done that, you can port/rewrite the patterns in C++.

Can you give those reasons please.

 

[QuantCoach]

Depends on the shop but coding might not be as emphasized as you think for a quant trader on a non D1 desk. The minority of quant traders that do actively code mainly use Python (and Excel).

 

[CrossGamma]

After all these different views, I am curious what @SixteenTick will tell us he actually used during his internship.

 

[Daniel Duffy]

Can you give those reasons please.

Well, OOP and inheritance is an optimisation step ... it is neither necessary not sufficient for successful software.
A more maintainable solution is Composition (for structure) and FP for behaviour.

OOP in Python seems to be an add on? (abc module) and is not a good implementation ... probably OOP is not really suitable for dynamically typed languages. It even supports Multiple Inheritance.

Encapsulation and Information Hiding are a challenge in Python.

 

[CrossGamma]

@SixteenTick - You should be a few weeks into your internship now. Please tell us what you ended up using.