Quant Research in Tokyo: How to Join the Machines
If you work in sales & trading, will you be replaced by a machine?
Maybe, but there are ways to reduce the risk of that happening.
One of the best is to learn to program and move into a quantitative role.
No matter how much artificial intelligence advances, computers probably won’t be able to program themselves effectively for quite some time.
Learn how to code, and you’ll be well-positioned for trading jobs, quant funds, and even tech companies.
Our reader today got into quant research a bit more randomly than that, so I wanted to get the full story straight from him:
Breaking In: Taking the Red Pill
Q: Can you walk us through your story?
A: Sure. I’m originally from a “European country,” and I initially studied accounting in university because I thought it was the most practical option.
But I became more interested in finance, did an internship in corporate banking, and then completed an equity research internship.
I liked the experience, but I was more drawn to quantitative analysis.
After that, I had the chance to visit Silicon Valley and see a lot of young people starting companies and launching technical projects.
So, I taught myself programming, did a few data science projects, and spent time applying my new programming skills by implementing and back-testing trading algorithms.
Then, even more randomly, I studied abroad in Japan, and after graduating, I asked the company I had previously interned at if I could return and work in the quant team.
They were hesitant since I didn’t have a math or science degree, but my side projects and experience in the previous internship convinced them to hire me.
I also got to stay in Tokyo, which has its advantages and disadvantages.
Q: That is quite a story, but I’m assuming that most people don’t get into quant research groups like that.
What types of candidates are these groups usually looking for?
A: The biggest myth about these roles is that you’re doing rocket science math or inventing new fields of math or physics.
But the work is much simpler than that: We mostly back-test algorithms and see how well strategies in research papers would have worked in past market conditions.
You need to know programming, and it helps to have a background in trading and finance; many people on our team have more traditional technical degrees.
Overall, large banks care more about your formal background for these roles, while asset management firms and independent trading firms don’t care as long as you can do the work.
I won my role mostly because of my side project, where I implemented a momentum trading algorithm described in a research paper.
The strategy had worked well in the U.S. and most of Asia, but not Japan, and I used my project to demonstrate the reasons why it didn’t work so well there.
Build a portfolio of projects like this, and you’ll go a long way toward signaling your skills and determination to quant groups.
Q: Which programming languages are the most useful for this work?
A: I used a lot of R and Python in school, as well as MATLAB.
My current boss is fairly open about languages because speed is not a big issue: We don’t do high-frequency trading, so we don’t need to use lower-level languages such as C to maximize speed.
Q: OK. And what did they ask about in interviews, considering that you had already worked there?
A: I spoke with the senior people on my team and the Chief Investment Officer of the firm.
Most of the questions were about data manipulation and SQL, and they threw in brainteasers and algorithmic problems.
They didn’t ask me to code anything, but they did show me an algorithm and asked me to find the bugs (it was incorporating biases in implementing a certain strategy).
They did not ask any of the usual “fit” questions about strengths/weaknesses and leadership, but that’s probably because I had interned there before.
Quant groups care far more about the technical side, but you’ll still get a few qualitative questions if you’re new to the firm.
Quant Research… in Tokyo and Around the World
Q: You mentioned there are advantages and disadvantages to being in Tokyo.
What are they? And what’s the industry there like?
A: The biggest advantage is that there are very few qualified, bilingual professionals here.
If you’re in a quant research group here, you will be contacted by a lot of headhunters recruiting for banks, tech firms, and other companies.
The biggest downside is that Tokyo isn’t the best place for this work: New York, London, and Hong Kong all have more positions and bigger teams.
If an asset management firm has a few thousand people worldwide, its Tokyo office might have only a few hundred, and only a small percentage will be in quant research.
The large international banks, Japanese banks, and asset management firms (e.g., BlackRock), all have quant research groups, but we focus on automating much of what Portfolio Managers (PMs) at traditional funds do.
Actively managed funds have struggled to justify their fees, given their performance, and clients often prefer automated strategies because they’re cheaper and easier to explain.
Q: OK. So, let’s say you’re trying to automate a long/short equity PM.
How would you approach this task?
A: We would start by looking at research papers written by firms such as MSCI, a leading tools provider for index and portfolio analysis. Large banks also send us many of these papers.
Then, we would implement the algorithm, apply it to historical data (“back-testing”), and see how it performs.
We do not spend much time generating brand-new ideas, but you might do more of that in the U.S. or Europe.
We don’t focus on new ideas because the Japanese market tends to lag behind developments in the rest of the world, so strategies from 4-5 years ago in the U.S. might be innovative here.
Once we get an algorithm working reasonably well, we spend a lot of time tweaking it and figuring out ways to rebalance portfolio holdings automatically.
Q: Why is the Japanese market so different?
A: One explanation is that “the mentality” is different: Right around the time Lehman collapsed and the financial crisis struck, the Japanese were still looking into subprime mortgages!
Major geopolitical events and crises make an impact here, but some people have argued that there’s more of a “cushion.”
Also, the economic picture has been quite stagnant for the past ~20 years, and aside from Abenomics, there hasn’t been much excitement in the markets.
Q: I see. What are the work environment and culture there like?
A: The quant research group is much less intense than the equity research group I worked in.
In an average day, I’ll start by back-testing some algorithms, tweak them a bit, and then show my results, along with some analysis of what worked and what didn’t work, to the PM or sales team.
That person will then show the results to the client, and the client might come back to us with some concerns – and I’ll have to make further tweaks.
But each cycle often lasts a week or more. It’s not like IB where everything must be done immediately and where you go through 113 turns of a presentation.
I don’t think I’ve ever stayed past 11 PM, and I usually leave much earlier than that.
Q: It sounds pretty nice.
Are you planning to stay there for the long term?
A: No, probably not. I’m satisfied with my current role, but I don’t think the asset management industry necessarily has a bright future.
We’ve seen a big squeeze on fees, and even with automated trading and algorithms, the lower fees will hurt us.
Plenty of recruiters have contacted me asking about my interest in tech companies, but I’m more interested in opportunities that combine tech and finance.
People in quant research tend to stay here for a long time because the money is good, the lifestyle is fairly relaxed, and, in a place like Tokyo, it’s tough to be fired.
Q: Great. Any other tips for students and professionals who are interested in this area?
A: I strongly recommend reading this primer on quant trading.
It’s geared toward automating strategies rather than quant research, but at least 50-60% of it applies to quant research as well.
There’s also a useful guide to back-testing algorithms here.
If you want to work in this area, do what I did and develop side projects implementing algorithms and trading strategies, get some experience at an existing finance firm in another area, and use your side projects to move into quant research.
Q: Thanks for your time!
A: My pleasure.
Free Exclusive Report: 57-page guide with the action plan you need to break into investment banking - how to tell your story, network, craft a winning resume, and dominate your interviews
Read below or Add a comment
Many thanks for the article and for the primer!
I’ve been hearing more and more about “quants” and how some of them make 250k-400k straight out of undergrad and how it’s so many people’s “dream job.” I’ve looked a bit at what a quant does and most of the time it’ll entail a ton of math, including stochastic calculus and Brownian motion. If I’m good at programming and good at math, then is this the career path I should be trying to go for? What are the downsides of being a quant? Why are quant guys the smartest people in Finance?
Yes, quants can earn a lot of money, but it’s not for everyone. Burnout/turnover rates are quite high, and it’s not that appealing to some because you never develop “soft skills” outside the pure math/programming, meaning that you wouldn’t be well-positioned to start or run a company afterward, move into management/leadership roles, etc. People often go into it and then switch out once they realize those downsides:
Thank you for response.
I am a European student interested in becoming a quant researcher.
In spite of having a heavy STEM background, I’m pretty decent with people and I’m a strong communicator. Hence, I really want to take on some type of leadership position later in my career. Do you have any insight into how that could work out? Basically, what’s the hierarchy like in the quant world? Considering that would you recommend me to pursue an quant researcher career? Or are there better options out there?
Anything’s possible, but if you want to do that, you need to move into more of a leadership role early on rather than remaining a quant for ~10 years (for example). So, I think you just have to do a few internships in these areas and see what you like. Not sure of the hierarchy at quant funds, but I’m fairly certain that even if you move up, you’ll still be doing mostly technical work (even PMs at non-quant hedge funds do a lot of investment analysis and reviews).
It’s a great article. I am currently an upcoming junior at the University of Rochester majoring in financial economics and minoring in comp sci. I am very interested in algorithmic/quantitative trading and want to work in a quant fund (or fintech) as a trader or in a role which is a combination of both finance and coding, not just coding. To pursue such a career path, what would be the optimal major choice.? I am sure about my financial econ major, but I am in a dilemma whether I should pursue a major in comp sci instead of a minor. As a minor, I do the extra work to learn and am currently working on algorithms to backtest strategies much as moving average and directional movement index. Would that major be a requirement to break into the industry because it sure will wear down my GPA a bit (which I can afford because I have a 3.8 and probably won’t go below a 3.6)? Also, if I want to work in quant funds, what companies should I be looking into considering I don’t go to a core school? The resources at my school are pretty good (for IB, S&T and consulting) but not for the industry I am trying to pursue because it’s very new. Are there any resources you recommend I check out?
I would appreciate any advice you could provide.
Probably something with statistics, math, and CS… you need to know how to code but you don’t need to be an expert for these roles. A minor in CS is probably better, and then major in finance or statistics or at least make sure you take classes in both. I can’t comment on quant funds to look at because I don’t know the industry well enough. Also, it’s almost impossible to win this type of role right out of undergrad, so it may be better to work at a large bank first and then move over… or go back to grad school first since many of these funds will require a higher degree.
Sort of along similar lines, is it possible to work at a fintech/A.I. company in a role that mainly involves coding/data science (or similar quantitative skills) but also requires a good understanding of finance right out of undergrad? If so, what would be the best major/minor combo? I know you have an article on fintech but I believe that it was geared towards business roles.
For background, I’m currently a freshman at Cornell majoring in Applied Economics and Management (AEM), and planning to concentrate in Finance and minor in CS. A third of AEM graduates go into investment banking and job placement is heavily skewed towards traditional banking/consulting paths. Because of this, I was originally planning on doing a 1-2 year MFE and then work as a quant at a fund, but I’ve become more attracted to companies where the final product primarily involves tech rather than generating returns. Is a higher degree still required for such a job that also pays well? Or does it depend on the specific company?
I appreciate any advice you might have.
Yes, it’s possible. You will probably need either a CS minor or major or something else technical to do it. You don’t need a Master’s degree to do engineering at a fintech company. If you want to spend most of your time coding, a CS major is better because you’ll be taken more seriously.
Hi Brian, thank you very much for this piece as this is what I need at the moment. I am grilled between two options – one is to go to JPM and do a semester internship in S&T, where I will be working in the electronic tradings’ product team. I hear that we would do something about the trading algorithms.
And another option is not doing the semester full-time internship and graduate on time for a return at a smaller bank’s IBD.
I am really interested in the quant research and my question is whether it would be a good start to work in electronic trading where we automate the trades if I want to move to quant research later?
If you’re more interested in trading, the S&T internship at JPM is better. IBD and S&T are completely different fields, so it really comes down to your personal interests. But if you want to do anything related to trading, quant research, or the public markets, starting in electronic trading and gaining some experience with automation is better.
Good interview. In reference to his comment about fees decreasing and more automation, do you think this might cause more “mom & pop” shops of 1 or 2 people running their own funds with small overhead?
i was thinking if you only have 1 or 2 heads to pay, small office space or even at home, you could probably survive the fee squeeze that is coming?
maybe instead of a few larger firms, you see more smaller firms?
In theory, yes, but the issue, as always, is raising enough capital to do it. Raising capital is always exceptionally difficult, and if anything, it’s harder for a mom-and-pop place. “Family offices” without LP money might become more common, though.
I am assuming that it’s only a coincidence but the WSJ had a piece on the topic last week. It comments on the finding of a NBER paper called “Replicating anomalies”. The authors reproduce 447 studies to conclude that more than 80% of the so-called anomalies disappear when more rigorous tests are applied. I don’t think the p-hacking skills of the industry surprise any insider anymore but I thought it was fun that some professors quantified it. That makes me wonder if the “Japanese delay” is a strength or a weakness; maybe the interviewee is wasting less time than me on the latest research fads.
The article can be found on the WSJ website (with paywall: https://www.wsj.com/articles/an-algorithm-an-etf-and-an-academic-study-walk-into-a-bar-1494528113) but the mentioned paper is free (http://www.nber.org/papers/w23394).
Anyway, thanks to both the interviewee and yourself. That’s an interview I wish I could have read before starting my current job. It wouldn’t have affected my choices but I would have understood my role better before starting!
Yes, the delay could definitely be a strength as well. Yes, I did see those links, but the timing was coincidental – I spoke with this person at the end of last year (have a huge backlog of interviews at the moment). Great series by the WSJ. Thanks for reading!
Don’t be so sure about the programming themselves thing, learning to learn is making decent progress (Bostrom et al. 2016) and ( arXiv:1606.04474 [cs.NE]) :).
I love everything about quantitative research. My only worry is that the work will be too repetitive. I’ve developed a little through StockSharp but settled on using QuantConnect for equities/options, MT4’s “Meta editor” for ForEx and Quandl databases.
For anyone else interested as well, if your broker is also TD Ameritrade you can dip your toes in the water with ThinkOrSwim’s ThinkScript (TypeScript essentially) platform and develope studies with alerts. To use their API you need to have a substantial amount of capital, so QuantConnect (Open Source) with its backtesting engine and MT4/ThinkorSwim for implementation should be enough to get anyone started.
My question for you, Brian or the guest writer, is this: Would you recommend quantitative research for someone who is actually VERY interested in research in all pure mathematics, computer science (specifically AI, data analysis, and machine learning, as that is my masters concentration)? I wouldn’t necessarily want to do something very relative (such as trading equities) and would want to have more of a “research” environment rather than a do whatever it takes to make it happen fast mentality.
Great article as always, and thank you in advance.
Thanks. I’m sure self-programming computers will happen one day. But it’s quite a bit harder than, say, self-driving cars or classifying photos.
I’m not sure that quantitative research in the finance industry is a great idea if you just want to spend your time researching pure math/CS – there will still be some team or client interaction in all of these roles, and in a lot of cases, you’ll spend more time on execution and testing than coming up with new ideas. It probably makes more sense to go to a big tech company with a large research budget in that case.
Would my math/science degree (sort of a weird story, getting an AS since I dropped out of high school my junior year to work full time in supply chain to afford some bills I had and after I earn this I have a transfer agreement to the state school here that’s relatively well known for there engineering comp sci) be a good fit for quant research along with all of the side projects I have? I’m only 19 now and have been doing this for a about a year but still even after that reply want to get into the quant research field. I intended to double major my bachelors with mathematics (PDEs and probability most likely) and computer science (computational science or machine learning/AI) then continue on with my masters in computer science. Would quantitative research still be a good idea? I only worry that if a quant fund were to see my background and notice I did not complete high school and initially attended community college, this would be a problem.
That may seem like a lot of rambling, but essentially my question is:
Being still only nineteen years of age, what would you recommend in my situation if I still wanted to break into the quant research field? I’ve actually started writing papers that I intended to attempt submitting to the arXiv, so perhaps this sort of thing may be of interest to them.
If anything, quant funds care less about pedigree than normal finance firms. If you want to break into quant research, get more finance experience… you need more than just math and CS because plenty of math people want to make a lot of money but have no interest in finance. Write algorithms to trade stocks, come up with new trading strategies, and be able to showcase them in interviews.
Great article! Thanks Brian! Love your work!
Thanks for reading!
Very insightful interview. Concise and straight to the point. I was just looking into what exactly roles in this part of the industry might entail, too. Great timing.
Thanks! Glad it was helpful. I’ve been working on the “concise” part…