“Quant firms” are finance companies that use mathematics and computers to guide investment decisions. Their key hires include people who are exceptionally strong in mathematics (the “Quants”) as well as those talented in computer science.
Our very own @hebegebe was a quant investor and lead researcher for 15 years, prior to his recent retirement. He has a strong understanding of the industry, including recent hiring patterns, and has offered to explain this field to current math and CS students.
Are you interested in a Math/CS major? Do you want to pursue a career in quant finance ? Make sure to ask your questions below!
College Confidential has vetted @hebegebe as an expert.
Are you a professional willing to share your journey and offer career guidance to our community?
Looking forward to learning about this field. Also interested in what coursework and skills are important in undergrad studies to get started (ie…the MUST haves) in this field and recommendations on coursework or skills (NICE to haves) that might set you apart from many applicants and get you noticed.
Thank you @CC_Sorin for the introduction, and glad to see some questions already.
I don’t know if there is an official definition, but here is how I think of it:
Quant: A person who heavily uses mathematics to guide an investment decision. The term “quant” is of course short for quantitative, meaning that decisions are made based upon measurable values as opposed to “gut feel”. More on that below.
Quant firm: An investment company that hires quants to direct their investment process.
The unsung heroes in these firms are the talented CS grads. While it is the quants that research ideas to make money, it is the CS folks that create the infrastructure to make that research possible, and then put the vetted ideas into practice. In a later post, I will be discussing the different roles in a quant firm.
But first, I want to provide a better description of the differences between quant investing and the more traditional type of fundamental investing. I will focus upon stocks because I understand that area best, but much of what I write below is applicable to other investment types, such as bonds and commodities.
Qualitative or Fundamental Investing
Examples of successful qualitative investors are Bill Ackman, Warren Buffett, and for those old enough to remember him, Peter Lynch. Successful qualitative investors often start out as analysts who become good stock pickers. They dive deep into a company to understand its role in the market, meet with its management to understand market differentiation and competitive threats, and estimate future revenues and profits. They use all this information to predict a future stock price.
Because these analysts dig so deep into a company, they are often called fundamental investors, and I will use that term here onwards to avoid confusion between qualitative and quantitative.
Fundamental investing is a labor-intensive approach, and therefore these analysts often focus their efforts on a particular sector (such as health care) and may have buy/sell recommendations on a few dozen companies. These recommendations are fed to a portfolio manager who takes recommendations from multiple analysts, each with their own specialty area, in order to create a well-diversified portfolio.
Disadvantages of Fundamental Investing
As I said above, there are certainly many successful fundamental investors. But there are also some disadvantages to fundamental investing. Some of them include:
Subject to emotion: Fundamental investors often meet with company management to get better insight into a company. This runs the risk of the investor being overly charmed by the management.
Subject to hidden risks: Portfolios created using fundamental investing can be subject to hidden risks. For example, if a portfolio is heavily weighted in both real estate and oil, they may look like very different sectors, but both are highly susceptible to rises in interest rates.
Labor intensive: As explained above
How Quantitative Investing is Different
Quants create mathematical models of expected future performance. One way they do this is by analyzing historical data and finding patterns that they believe are likely to persist over time. This data can include information about the company’s earnings, its price movement history, textual analysis of changes to wording in annual financial statements, and many other things.
Once a quant finds patterns that they believe will continue to predict stock prices (how this is done properly will be described in a future post), it is vetted by a management team before it can be made part of the investment process. That’s when the CS gurus take over. It is their job to take a good idea and implement it in a way that both works reliably whenever possible, and perhaps more important, fails in a safe way when it cannot work (perhaps due to lack of sufficient data).
Some of the advantages of quant investing over fundamental investing include:
Scale: Once the right algorithms have been developed, they can be applied to thousands of stocks, across many industries.
Rational decision making: Because investment decisions are guided by algorithms, the pitfalls of emotional decision making are kept in check
Risk controls: Investment decisions automatically take risks into account. For example, it would have a threshold on how much interest rate risk is acceptable, preventing a portfolio from being overloaded in energy and housing stocks.
Thank you for taking the time to answer questions here!
Could you give an example of a quant hypothesis that one might test or put into production? Do you have any that you were enamored with (or does that go against the very principle of the concept?!) that ended up being incorrect?
I think of “finance” and I think of a culture like The Wolf of Wall Street / Gordon Gekko. That doesn’t really square with what I think of with CS and math majors. What is the culture actually like in quant firms? Are there firms that have “nicer” reputations?
Education-wise, what does the process look like? I assume there’s some sort of a pipeline that you have to get into in order to get internships, etc.? Is an MFE necessary, or how does one go about getting a job in the industry?
I am skipping a few questions and answering this one first because this one helps people understand the job market in these companies.
But before we get into roles, we first need a description of the different types of companies that hire quants and their CS counterparts.
Quant investment firms: These are investment companies that primarily make investment decisions using an algorithmic process. They buy assets, typically hold it for some length of time ranging from months to years, and hopefully sell them for a profit. They also frequently “short” assets as well, in which case they hope the price goes down.
Fundamental investment firms: Fundamental investment firms will hire a small number of quants for risk control. While an individual fundamental portfolio manager might not consider risk very carefully, it’s essential for the firm to do so across all their portfolio managers.
Trading firms: The purpose of these firms is to create a market. If you think of investment firms as being equivalent to art collectors, then the trading firms are the art dealers. They buy stocks they believe is underpriced and sell what they believe is overpriced. Depending upon the firm, their holding time can range from milliseconds to a few days. Note that both quant investment firms and fundamental investment firms work with trading firms.
Note that there are also firms that primarily focus upon creating low-cost indexes, such as Vanguard. They also have a risk group, but it’s small by comparison even compared to the fundamental investment firms.
Now, let’s look at the names of quant investment firms and quant trading firms, as they are less well-known than fundamental investment firms:
Quant investment firms: Some well-known companies that primarily invest quantitatively include Acadian Asset Management, AQR, ArrowStreet, Bridgewater, DE Shaw, Numeric Renaissance Technologies, Two Sigma and Wellington. There are also quantitative divisions in large companies like Fidelity and Citadel. This is only scratching the surface as there are hundreds of others.
Trading Firms: Well known examples include DRW, Hudson River Trading, IMC, Jane Street, Jump Trading, Optiver, SIG and Virtu.
In terms of roles, I will start with trading firms because they have the most breadth, and it’s a progression from being purely math based to being purely CS based.
Traders: In a trading firm, these are the rainmakers. In companies like Jane Street and Optiver, these people take rapidly incoming data, use math to quickly estimate how it will affect prices over the short term, and direct trading decisions that could be tens of millions of dollars. They need the ability to shake off losses and make the next decision with confidence. Little to no programming skill is required, and the key skill is similar to that of good poker players, and in fact Jane Street hired the best woman poker player: Vanessa Selbst, female poker player, quietly works for Jane Street
Quant Researchers: These are people who look at data more carefully, perhaps taking weeks or longer to investigate a particular area. For example, someone might be looking at how the evolving war in Ukraine is likely to affect grain prices throughout the world. The result of their research feeds in as guidance to the traders that are making the real-time decisions. People in this role are expected to have a mix of both strong math skills and programming skills.
Software developers: No specialized statistical knowledge is required for these positions, but in addition to top-notch development skill, this role requires extreme attention to detail. Investment firms move a great deal of money, and errors can quickly sink a firm.
Risk modelers: While it’s the traders that make the money, it is the risk modelers that keep the firm from losing too much money at once. The skills for this position are similar to that of the Quant Researchers.
Note that some trading firms (such as HRT) do not have people making trading decisions, but instead do everything through computers. There the role of the Quant Researcher is given more importance.
Quant investment firms give great value to the roles of Quant Researcher, Software Developer, and Risk Modeler. They have traders as well, but they have little discretion and therefore are not highly valued.
The fundamental investment firms and index investment firms give great importance to the Risk Modeler, but the number of people hired for this role in each company are small. The CS people in these companies do important work, but are not as valued here as in the quant investment firms and quant trading firms.
Before we get into the requirements, let’s talk about what type of person is a good fit for these jobs. Were you, or your child, someone who just loved math/programming as a kid? Someone who did math/programming problems for fun? Thought of math competitions/hackathons as something to look forward to? If so, then you may be a good fit for a quant firm.
Yes, these positions pay well. But I think the people that do best in these roles are the ones who love doing the work and think “I can’t believe I get paid to do this.”
Regarding the requirements, let’s look at it terms of the different roles:
For trading roles, it’s all about being able to evaluate probabilities quickly. Courses that may help include Probability, Combinatorics, and Linear Algebra. And learning to be a really good poker player (only half joking here).
For research roles, combining statistics with computer science is ideal. In addition to the above listed math courses, students should take courses in linear modeling, hypothesis testing, artificial intelligence, algorithms & data structures. If possible, take graduate courses in these areas. They should also have a strong understanding of Python and the libraries used for data analysis (I’ve stepped away from this a couple of years now, but back then the main libraries we used Numpy, Pandas, PyTorch and TensorFlow).
For CS roles, more than anything else it is about algorithmic efficiency. Pay close attention to the Data Structures & Algorithms class. Practice leetcode until you can do mediums with ease and get some hards in time.
In addition to the above, external validation also plays a role for the most selective companies. Examples of external validation are doing very well in competitions in high school (e.g. USAMO, USACO) or college (Putnam, hackathons, etc.).
Depends upon the role. Trading firms readily hire traders straight out of undergrad. CS jobs are also readily hired right out of undergrad.
For research positions, it depends upon the company. Some only hire PhDs, while others will look at students with masters degrees, and a smaller number will look at students with undergrad if they are comfortable with the student’s knowledge of statistics.
I can give one example of where the conventional wisdom suggested one thing and the research said another.
The research topic was about diversity in the boardroom. There were many news articles stating that companies with diverse boards outperform those without them. We wanted to find out if there was a causal relationship or if it was just a correlation. Because if it’s a causal relationship, then we can use it for stock selection. But if not, we can just ignore it.
We found a data source that had a historical board diversity score (which if I remember correctly was effectively the percentage of the board that wasn’t white men) for large public companies. What I did was look through the data and found when the diversity score increased, and looked at stock returns of that company before and after that date, relative to others in that same industry.
What we found is that the relative returns of the company were good after the diversity increase, but they were even better before the diversity score increase. Simplifying, before the diversity score increase these companies were firing on all cylinders, and looked around at what else to improve, saw that the board wasn’t very diverse, and fixed that. And it wasn’t that the diverse board made the company less successful, but rather that even good companies cannot maintain outstanding stock performance for very long. In the end, we just ignored the board diversity score.
On the positive side, these firms are filled with bright people and tend to have top-notch equipment. But that’s separate from culture. And culture in these firms varies a great deal, just like tech company cultures vary a great deal.
Companies like Jane Street and Two Sigma were often compared to Google (before Google’s recent layoffs) in that they have bright people working in a collegial atmosphere, and most employees stay long term. Typically averaging ~50 hours per week.
Companies like Optiver, Citadel and SIG are known more as “sink or swim”. They give a certain amount of time for people to perform, and if they don’t they are out. But they are also known for paying their best performers even more. Because of the pressure to perform, hours can be intense at times.
As I wrote above, the education requirements vary by role.
Also, I didn’t mention anything about finance knowledge earlier. A lot of the most selective companies don’t care about finance knowledge at all. Their view is that they can teach finance knowledge more easily than they can teach math/CS talent. Examples that don’t care at all include: Arrowstreet, Citadel, DE Shaw, DRW, HRT, IMC, Jane Street, Jump Trading, Optiver, SIG, Two Sigma, Virtu.
On the other hand, companies like Fidelity, Numeric and Wellington do expect finance knowledge.
In terms of pipeline, some schools are targets. They include the usual suspects like HYPSM. But they also readily hire from colleges like CMU, Michigan, UT-Austin, UIUC although they may favor students from certain for specific roles. An interesting one is Baruch College, part of the City University of New York. It has a well-regarded trading center, and some students from there do land coveted trading spots.
Note that it’s not just about GPA. I know a Yale CS student with a 4.0 that was summarily rejected from just about every trading job he applied to, likely because he didn’t show the math background they were looking for. Likewise, I also know of students with lower GPAs that showed math ability through awards that were readily called in.
Most companies begin their process with an online assessment that is tailored for the specific role. Many companies send this out to anyone that applies, regardless of any other qualifications. Doing well on this is necessary for the companies to actually start considering someone as a candidate, but it doesn’t guarantee any further progress.
I don’t have a good answer for the AI question. Our firm was behind on AI, but performed well despite not using it much. Our approach was that we started with a hypothesis as to why something would be true, and then we tried to confirm that hypothesis through backtesting. We never resolved how to properly integrate that approach with AI. But I know that other firms used AI much more extensively than we did.
It’s unfortunately very low, perhaps 10% or so. Many firms would like to hire more.
Hi — thanks for doing this AMA. I have a question to ask. How does the interview process differ among the various groups of these companies — eg 2 sigma and DE Shaw vs JS, 5R, and SIG (ie the SIG derivatives) vs the Europeans (optiver, imc etc) vs the smaller guys (drw, arrow street etc). If you think the groupings are wrong, please correct me. I assume Virtu is sort of different from the rest of them? Specifically what type of questions are asked etc. for QT and QR. Which shops are C++ heavy for the interview even for QT/QR. Thank you.
I don’t have detailed information about each company’s interview process. But here are some hopefully useful nuggets:
The quant companies spend a lot of time interviewing, even for interns. By the time an intern gets to a final round, it’s pretty typical to have an online assessment plus anywhere from six to ten one-on-one interviews. Given all the rejects along the way, they are probably spending 60-100 hours per intern that joins them for a summer. Finding talent is extremely important for them.
Jane Street is known for having the most organized interview process. There are 3-4 rounds, and within a few days they will tell if you have passed to the next round, which is typically scheduled within a couple of weeks. So a person can go from start to an offer in around 6-7 weeks. For the final round, students are invited to New York for an all-day event.
SIG does something unusual with offers. After a student is done interviewing with them, they don’t tell them if they got an offer or not. Instead, students can decide when they want to hear if they got an offer, and SIG tells them at that time. If they get an offer, they have a week to accept. I think what SIG has a target of how many hires they want, and among the candidates they liked, they keep offering 1-week exploding offers until they fill up.
@hebegebe
I know the default coursework recommendation is the most rigorous math possible, but at some schools that can shoehorn students into highly abstract real analysis/abstract algebra/topology courses. Would you recommend them over the more applied/relevant alternatives for signaling reasons above? At what stage in the interview process does one’s transcript get looked at?
For quant trading roles, a strong understanding of probability is extremely important. Depending upon the company, required software skills can range from non-existent (Jane Street) to really quite strong (HRT).
For research roles, students are not penalized for taking a theoretical math curriculum as long as they also have some classes on linear models, AI, and a strong understanding of statistical methods. Note that research roles are relatively rare for undergrad students.
I expect that every company looks at the transcript after a student completes the online assessment (if the company has one) and before the first interview.
As in a calc-based probability course or a measure-theoretic probability course?
I know some companies (which ones?) consider IMO/IOI performance to be a plus (is that for trading or research roles?), but what about other olympiads like IPhO, IChO, IOAA?