AMA: Former Quant Turned Data Scientist

[ibpkpnu]

Per suggestion from @Andy Nguyen, this is a follow up on this thread.

I attended a top MFE program in the US and currently work as a data scientist at a tech unicorn. I previously worked as a researcher at an asset manager on quantitative equity and systematic global macro strategies and than at an equity and futures statistical arbitrage hedge fund. Currently my job includes model/algorithm research and development and writing production code drawing on methods in econometrics and machine learning.

I am happy to answer questions on my experience and tech vs finance careers.

 

[marcusaurelius]

Thanks for making this AMA; it is appreciated and helpful to us younglings!

I took the time to read the other thread beforehand. For context, I am myself entering CMU's MSCF program this August.

1. I must admit that your story speaks to me. You mentioned in the other thread that the tech industry is a better fit for you, when comparing for stress levels, risk, and volatility. You also mentioned that, in retrospect, you probably would not be pursuing an MFE. If you were in my shoes, and wanted to keep the tech data analysis route open, what skills would you try emphasizing to maximize outcomes for both quant finance and tech? I believe you mentioned emphasizing optimization/stats/CS over stochastic calculus/PDE/derivatives pricing: could you go more in-depth (perhaps textbooks you particularly like, or other resources?) On that note, how would you go about showcasing these skills efficiently? Having a personal website was mentioned in the other thread; other ideas on what you would like to see should you be recruiting someone?

Just want to learn from your story, as I am perhaps quite close to where you were nine years ago!

For context: I am myself closer to the macro side with an econometrics background.

Many thanks in advance

 

[weab]

Hi @ibpkpnu , on your last thread you mentioned that top quant funds do not hire from MFE programs. Is this because much of the MFE curriculum focuses on sell side derivatives pricing or top quant funds do not respect the degree in general? I did my undergrad in pure math and economics. If I take rigorous classes in math, statistics, programming, and machine learning as part of my MFE will top buyside firms actually consider me?

Thank you

 

[zephyrus]

Hi!
You mentioned in your last thread

Many MFE programs traditionally focus on stochastic calculus, PDE, and derivatives pricing, skills for which the demand has been stagnant at best for the last ~15 years. Core math, statistics, and computer science knowledge is durable over time and portable to many different jobs

Would a masters in CS/statistics be a better (and safer) choice for quant industry acc to you?

 

[ibpkpnu]

Thanks for making this AMA; it is appreciated and helpful to us younglings!

I took the time to read the other thread beforehand. For context, I am myself entering CMU's MSCF program this August.

1. I must admit that your story speaks to me. You mentioned in the other thread that the tech industry is a better fit for you, when comparing for stress levels, risk, and volatility. You also mentioned that, in retrospect, you probably would not be pursuing an MFE. If you were in my shoes, and wanted to keep the tech data analysis route open, what skills would you try emphasizing to maximize outcomes for both quant finance and tech? I believe you mentioned emphasizing optimization/stats/CS over stochastic calculus/PDE/derivatives pricing: could you go more in-depth (perhaps textbooks you particularly like, or other resources?) On that note, how would you go about showcasing these skills efficiently? Having a personal website was mentioned in the other thread; other ideas on what you would like to see should you be recruiting someone?

Just want to learn from your story, as I am perhaps quite close to where you were nine years ago!

For context: I am myself closer to the macro side with an econometrics background.

Many thanks in advance

I would emphasize the following items to maximize outcomes or maintain optionality for both quant finance and tech jobs:

1. If you are working in finance, work at a place where technology is a first-class citizen and strategic priority. Many of not most firms in finance do not invest in software the way a Google, Meta, or venture-funded tech startup would and run the business on old, legacy software with weaker development standards. Investment banks are bureaucratic, highly regulated, and would fall in this group as would many large asset managers. Many HFTs and quant hedge funds are much stronger with tech and employ many people from FAANG+ tech companies (and also lose people to these tech companies). Firms like Jump, HRT, Citadel, Jane Street, Two Sigma would be in this group. Your employability in tech will suffer if the programming languages and software frameworks you use in your job are far from the cutting edge.

2. Work on projects which advance and demonstrate your skills in core, more general data science and machine learning and data analysis applications. Pricing derivatives using PDE or doing classic factor investing is unlikely to help much here. There are increasing applications of machine learning (including deep learning and reinforcement learning) in finance including the research of Bryan Kelly, Marcos Lopez de Prado, Gordon Ritter, and Blanka Horvath. I also follow Vivek Viswanathan on LinkedIn and he does a good job outlining his evolution from a traditional factor quant to an ML quant while stressing the importance of solid software development and data management practices. Many finance firms will have teams focused on data science and machine learning which operate in a similar way to those at tech companies and can be great places to work. With the proliferation of new, "alternative" data sources also comes a need to handle less structured data, e.g. text, in a way which is most suited for ML techniques.

3. Write production code. I have interviewed many people for both quant and data science jobs, and a poor grasp of CS fundamentals and lack of programming experience is the most common deficit I see. Many quant and DS jobs do not require you to write production code, but your impact and opportunities will be much higher if you can.

While investing and trading is fundamentally about liquidity provision and/or managing a portfolio, tech is (mostly) about building a tangible product. It is hard to develop tech product intuition working in finance, but you can build this on the job in tech.

 

[ibpkpnu]

Hi @ibpkpnu , on your last thread you mentioned that top quant funds do not hire from MFE programs. Is this because much of the MFE curriculum focuses on sell side derivatives pricing or top quant funds do not respect the degree in general? I did my undergrad in pure math and economics. If I take rigorous classes in math, statistics, programming, and machine learning as part of my MFE will top buyside firms actually consider me?

Thank you

Some top firms do hire at MFE programs, but I think it is one among many different degree programs they consider and individual talent/performance matters a lot. Some firms have a strong bias for PhD graduates and the MFE is a masters degree. At my last job in finance I was the only MFE and one of the few non-PhD researchers, while in a previous job there were several MFEs. Others will hire from all degree levels (including bachelor's) and focus hiring mostly at elite schools (as is common in non-quant finance). You can look at the placement records for programs to see where students were placed in jobs.

If you take rigorous classes in your program and do well you should get interviews at quality firms. For buy side jobs I would focus more on econometrics, statistics, optimization, and ML and less on PDE and stochastic calculus. Career offices of MFE programs also function as sorts of recruiting agencies and will have a major impact in what interviews you are able to get. There is considerable variance among programs in their abilities to secure interviews and place students in jobs.

 

[ibpkpnu]

Hi!
You mentioned in your last thread

Would a masters in CS/statistics be a better (and safer) choice for quant industry acc to you?

CS/stats would give you more optionality outside of finance (e.g. in tech), but I don't think it is necessarily a better or safer choice for quant jobs. The MFE curriculum is designed specifically for quantitative finance and the career offices are focused on placing students in quant jobs. A downside of the MFE is that it is relatively narrow and perhaps overly vocational, while CS and stats emphasize more fundamental, general knowledge.

 

[marcusaurelius]

I would emphasize the following items to maximize outcomes or maintain optionality for both quant finance and tech jobs:

1. If you are working in finance, work at a place where technology is a first-class citizen and strategic priority. Many of not most firms in finance do not invest in software the way a Google, Meta, or venture-funded tech startup would and run the business on old, legacy software with weaker development standards. Investment banks are bureaucratic, highly regulated, and would fall in this group as would many large asset managers. Many HFTs and quant hedge funds are much stronger with tech and employ many people from FAANG+ tech companies (and also lose people to these tech companies). Firms like Jump, HRT, Citadel, Jane Street, Two Sigma would be in this group. Your employability in tech will suffer if the programming languages and software frameworks you use in your job are far from the cutting edge.

2. Work on projects which advance and demonstrate your skills in core, more general data science and machine learning and data analysis applications. Pricing derivatives using PDE or doing classic factor investing is unlikely to help much here. There are increasing applications of machine learning (including deep learning and reinforcement learning) in finance including the research of Bryan Kelly, Marcos Lopez de Prado, Gordon Ritter, and Blanka Horvath. I also follow Vivek Viswanathan on LinkedIn and he does a good job outlining his evolution from a traditional factor quant to an ML quant while stressing the importance of solid software development and data management practices. Many finance firms will have teams focused on data science and machine learning which operate in a similar way to those at tech companies and can be great places to work. With the proliferation of new, "alternative" data sources also comes a need to handle less structured data, e.g. text, in a way which is most suited for ML techniques.

3. Write production code. I have interviewed many people for both quant and data science jobs, and a poor grasp of CS fundamentals and lack of programming experience is the most common deficit I see. Many quant and DS jobs do not require you to write production code, but your impact and opportunities will be much higher if you can.

While investing and trading is fundamentally about liquidity provision and/or managing a portfolio, tech is (mostly) about building a tangible product. It is hard to develop tech product intuition working in finance, but you can build this on the job in tech.

Valuable info. Much appreciated! @ibpkpnu