Hey Everyone! I’m Rohan. I currently work on AI ranking and optimization at LinkedIn but I used to work in a some trading firms/brokerages in areas like portfolio optimization, smart beta etf optimization and some order execution stuff (mainly tca stuff).
Haven’t been in quant for a while but there’s always something which has always drawn me to the type of problems - probably the wide array of interesting optimization problems and risk profile constraints.
I’ve learned a lot from the channel and would love to continue learning here.
Hi Rohan, nice to meet you and thanks for sharing your story.
AI at LI must be very interesting. Absolutely tons of data to make AI actually work.
Finance and AI are a bit of a tricky one I think. I think AI is best when it’s very obvious but very tedious for a human to do the task, but in finance we are asking the machine to be even better than the human. Not yet for sure!
Very happy to hear you like the channel. What sort of content would you like to see more of?
Do you have any recommendations for your quants wanting to get into your field?
I’ve read a decent bit of Igor Halperin’s book on ML in Finance - I think it’s great and he covers some great tools for financial applications of ML and also one of the only books I’ve seen with a whole section on explainability - learning a lot of what is and isn’t possible.
Personally I’ve liked the fact that the channel has focused on more tools for validating assumptions and metrics for portfolios. It would be great to hear some of your thoughts on parameter estimation and things like shrinkage estimators for portfolio optimization - or just some topics you find interesting in portfolio optimization in general. You have some great experience so it would be cool to hear how you deal with evaluating market impact, bad order execution, and trading costs effects on your strategy.
I will have to check that link and that book out. There is always something new to learn in this space.
I gotta day, I am much more of a fan of the classic methods, which is very likely a bias of mine because I feel more comfortable in that space, but there is undoubtedly lots of value to be extracted from ML.
I am thinking of REAL life applications, like missing and spurious data. People think being a quant is all about using models etc. Not really. The old agade still holds true, about 80% data wrestling, munging and cleaning, and 20% corporate admin lol.