I've seen a lot of what I would consider to be misinformation on this forum about what being a quant is like, where is good to work, what is needed to get in, etc etc. I wanted to give my personal thoughts on what I believe to be fact vs. what is fiction.
Why you should listen to what I have to say:
First off, you dont have to. That being said, I do think I am able to provide a perspective that is a bit unique on this forum. During undergrad I was successful during IB recruiting for my junior summer, and landed in a group that most people on this website would consider "ok." During my internship I decided I didn't like the people, and turned it down my return w/ nothing in hand.
My FT recruiting was initially split between RX and Quant. My RX pipeline was looking MUCH better than the latter, but I decided the barrier to entry there was very low, and if I wanted to do Quant I had to do it straight out of school.
I now currently sit in a MM pod (multi manager, not market maker) as "Quant Researcher." I am the first quant to ever come from my school. Having done both Banking and Quant, and recruited for both I think I can help clear up some of the misconceptions that both parties may have.
All this being said, I am far from a senior employee (have been working 2-3 years) and likely am still wrong about a lot fo things. If you want better advice / info, go read posts from these guys:
MMPM, [Tobin's-Q], anonq
I'd love to invite their commentary as well (not all are quant but have good perceptive / have been helpfull imo)
What everything I am going to say applies to:
Everything I say regarding the quant role only applies to fully systematic (no human intervention) quant research. I don't have a valid opinion on Quant Traders, Quantamental, Market Making, Prop, etc because I dont have the firsthand experience. I could tell you what I have heard / been told but this is often much noisier and lower value information.
Common misconceptions / questions:
Do I need a PhD to be in a research role? - No, absolutely not. I use heavy duty ML (parralelize stuff across 2000+ cores every day to give an idea of compute), read lots of papers, take part in research meetings, etc. I have an undergrad math degree and no relevant internships. This isn't to say you don't have to be technically proficient, as you absolutely do, but this whole notion that you need a Master/PhD/IMO Gold to become a researcher is absurd.
With this comes hot take number one: The average undergrad quant is likely to have better outcomes than the average PhD quant. Part of this is due to selection - PhD quants are often people who couldn't cut it in academia / their chosen field, whereas the undergrads doing this often really want to be here. Smart undergrads also often know they dont know shit and as a result arn't impaired by arrogance, which plagues PhDs.
Do I need to know how to code? - Absolutely yes, but not necessarily in the way you think. If you are gunning for larger shops with a defined recruiting pipeline, you're going to need to leetcode and there is now way around it. However, if you are going to somewhere without a defined pipeline, having projects that demonstrate you know how to manipulate data efficiently (and being able to reproduce these techniques on the spot) will be sufficient in most cases. You will spend more time coding on the job than anything else - its the language you do your work in. An appropriate analogy would be asking an author if they need to know how to type.
What math do I need to know? - This is one where it's more specific to how you trade and where. If you are a sell side boi where you need to do stuff with some explainability and academic grounding, then stochastic calculus will be needed. However, if instead you are focused on make strong predictions and want to make money, then the math needed to understand popular ML techniques is sufficient. This is a vague description, but its the right one. Just like the coding, the math is a tool to make money. You need to understand it well enough to work through and troubleshoot any models you build / use them properly and in a way that makes sense. I have sat down and solved a "hard" math problem exactly one (1) time in my career, and it didn't end up making any money. For people who really want a list, here: Linear Algebra, Differential Calc, Probability Theory, and basic Calc Based Stats (none of that business school garbage).
Is networking important? - Similar to banking, it is one of the most important parts. This is likely a bit of a hot take, so I'll elaborate. Quant finance is a very small world, much smaller than the fundamental L/S one where it already seems like everyone knows each other. While the first gig may come from one of those big pipelines that is impersonal and regimented, every good opportunity comes from knowing a guy who knows a guy. You know how everyone says in the HF world that all the best opps come from referrals / never get posted (Tiger Cubs, etc)? It's the same shit in quant. Similarly, there is a HUGE world of opps for new grads outside of the big bureaucratic firms who's names you recognize,, and you only learn about those by digging around and emailing.
Where is a good place to work? - This is entirely up to the individual and depends on a lot of factors. If you care about brand and plan on exiting to SWE in a few years / dont really care about finance, go work for 2sig. But if you are interested in finance, care about knowing the whole process, and wan to do this kind of work for the long haul, I would strongly recommend targeting small teams that manages their whole process (idea generation all the way to execution). And with this, comes another hot take: The best place to start a career as a quant is in a multi manager pod. This is obviously dependent on having a PM who cares about your development and knows how to make money, but this had huge advantages. A user on this site put why this is the case better than I could (replace IB Stat Arb with Multi Manager Pod):
"I think the Stat Arb track at IB's was good because it was small teams that had to do everything. As opposed to now a days if you start at like Two Sigma/Squarepoint/etc then you only see a small piece of the pie. I'll interview people from places like that and the only thing they ever worked on was portfolio optimization or some just an alpha or two on a really niche dataset."
This is one of the points that I feel most strongly about from my limited experience. The larger the shop your work at, the more silod you become. The average junior at a Multi Manager / Prop Shop / 10 man fund is going to have a hell of a better idea on how to actually make money than almost any junior at 2sig. If you want to run a book someday or be responsible for PnL, I feel this is critical. The difference between vertical siloing (many teams run their own full process independently) and horizontal (many teams each manage one step in the process on their own) is night and day.
What are the exits for quant? - Ask a L/S HF guy what the exits are from their job. It is the exit. If you want a well defined path where you always know where you'll be at T+2yrs, dont be a quant. I would discourage anyone who wants to jump quick from entering this space (unless you flopped your google interview and want to try 2sig). I understand people want some security in case things go belly up / they get forced out of the industry, and to these people I would say industry data science / fintech I guess? It isn't something I think about much and dont ask people either. The best plan B is no plan B.
Since people love recruiting questions, heres a few quick ones:
When does recruiting happen? - Late summer though the end of Fall
How should I apply to firms? - Amass a long lost of companies you would work for. Group them into 3 ranks, 1 - Dream company, 2 - Great company, 3 - Would work for them as a first job. In three "waves" throughout recruiting season, apply to a few companies in each group. This ensures that you get interview practice at each "level" of firm without burning all your chances with dram companies you just weren't ready for / firms that would be layups once you get practice in the start of the recruiting season. It gives you time to practice live, which is key for quant interviews. Remember, you only need to win once. I'm a fan of being pretty polarizing in interviews / avoiding being meek or boring. The median candidate never gets the offer, so sometimes you're better off turning the continuous distribution into a coin flip by being a bit different.
How should I prep for interviews? - Leetcode, brainteasers, mental math. Leetcode has been beaten to death on how to do it, just google how to study. For brainteasers, look at the green book and heard on the street. Those cover just about everything and whatever you get will likely be some subtle change to one of those - focus more on understanding the thought process behind an answer vs. knowing the answer, as this lets you generalize when you get a curve ball. For mental math, look up Arthur Benjamins book and drill that.
How can I make sure I get a first round? - I had a 100% hit rate for getting the screener by emailing recruiters at the firm / traders asking to talk to them. Applying IB recruiting techniques to non-IB industries often gives absolutely killer results.
I hope this helps - people on this forum have been great to me, and I wanted to give back in some way. More than happy to answer any questions below, and would love to see some of the quants on here provide their perspective.
Thanks for the details... I have a few questions:
1. Does your school really matter? I have always heard that Quant firms primarily recruit from certain colleges?
2. What about experience? While students do not have usually have much, does doing algo trading on the side help with the passions/why Quant question?
3. Lastly, how is WLB? Has it generally been 40-60 hours, no weekends?
Thanks!
1. Does your school really matter? I have always heard that Quant firms primarily recruit from certain colleges?
It does, but it isn't the end all be all. If you go to HYPS or MIT, you will almost alway get the benefit of the doubt. Because this field is a bit more skills based (as opposed to banking), being from a non-target hurts less than in banking. You see plenty of non-targets crush it in banking, and you'll see the same in quant. While it'll be harder to be in a position to prove passion/competence, once you get that it's mostly even. Similar to all other fields, if the kid from harvard is bad no one will blame the guy who hired him. If the kid from Rutgers is bad, people will thinking "well what did you expect?"
2. What about experience? While students do not have usually have much, does doing algo trading on the side help with the passions/why Quant question?
I had actively developed my own strats on the side. They had many, many problems but in interviews I directed the focus on what these problems were, how I found them, and how I think they could be fixed. I think this was a huge differentiator - there was no shortage of kids claiming to have a sharpe 10 algo that printed money on thier resume. I talked about a sharpe 2 model that I had hours of talking points of improvements for. Self awareness is important.
3. Lastly, how is WLB? Has it generally been 40-60 hours, no weekends?
I don't know any quants who have killed it long term who dont work like dogs. Right now I work a bit every Sat/Sun, and probably 7-8am to 10pm most nights. If you want to get ahead in any field, you need to aggressively put time in at the start. Quant isn't an exception - this is something I should have actualy put in the above post. The one caveat is like most buy side jobs, I have a much higher degree of predictability in my schedule - by construction there wont be any surprised on weekends
Thank you for contributing to the community. May you answer the followings:
What are some firms/groups you would recommend? Also, is there any tips on grinding the technicals other than go for it?
What are some firms/groups you would recommend?
I wont give specific firm names, but I'm happy to give some things to look out for. As with all my advice, the only thing I feel comfortable speaking to is fully systemic, realtivly high capacity strategies. In general, high tenure among all personnel is a good sign, ideal with the people progressing in responsibility as time goes on. It can be hard to judge the last part of this because titles are generally "Quantitative Research" or "Portfolio Manager" - only real way to get a read is to talk to people. Similarly, if you're looking at a MM pod, and the whole team has been there for 5+ years, that typically means its a very good place to be. Additionally, AUM (or GMV if in that kind of setup) per head can be a good proxy for how much money there is to go around. Really the same metrics that apply to any other fund / group apply to quant pods as well - many principles of the money management business transcend investment styles.
Also, is there any tips on grinding the technicals other than go for it?
The hands down number one rule for grinding any technical question is to NOT LOOK AT THE ANSWER. This is something I've told many people in the past, and they dont do it and it kills me. For every question you do, do not so much as peak at the answer or any hints until you have worked through a solution that you are very confident is correct / thought through any edge cases / are sure you have given your best effort. Then, once you have this answer, check ONLY IF IT IS CORRECT OR NOT. Don't look at any process details / proofs / explanations etc - you either were 100% right (no "sort of or mostly") or wrong. If right, then congrats, read the explanation and move on. If you were wrong, look at nothing, go back to the problem and work it out again. If you go through all the leetcode / brainteasers like this and are disciplined, you will absolutely crush any interviews because you will have actually learned how to think and solve these kinds of problems. If you dont, you'll get a false sense of confidence and crash and burn come interview time. Doing it right sucks, and sometimes you can get stuck on a problem for hours / even a day. Its incredibly important that every answer you arrive at is done solely by your own deductive skills and not any other aids. Prepping in any other manner is effectively introducing a lookahead bias into your interviewing - it'll look great in the backtest, but will break down in spectacular fashion as soon as you go live.
Do you have any strategies for "digging around" and finding smaller firms to apply to? My friend got an internship at one without any technical questions.
This is a hard thing to master, but very important for how I think is the best way to go about landing your first gig.
Multi Manager Pods:
First, get a list of multi managers - there are a LOT more than you think, and many that arn't as popular overall do very well for quant. For example, millennium has a great product and everyone knows who they are, but from what I've heard their quant setup is effectively "here's a server now go fuck yourself." All multi-strategy funds have a favorite - find out where that is quant. Once you have a list, start reaching out to PMs. Be smart when you do this and dont burn bridges.
Single Mangers:
Similar to how most MMs have a favorite strategy, many allocated do as well. Dig around for these, and figure out who is getting allocations that you haven't heard of. There's dozens of shops that are great to work at that you've never heard of. Unfortunately, the best way to find these is to just be interested in the space and always digging. One thread that names drops a good number of firms is here (https://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=4851) (if you don't find going through all pages of this thread interesting, maybe rethink the quant stuff)
One thing to remember is that similarly to how it is hard for you to find these "no name" firms, it's just as hard for them to find competent, enthusiastic talent. If that gets dropped at their doorstep, that makes thing a lot easier for all parties.
What are some concepts you should learn at school? Besides python, would you recommended learning stochastic/regression/or more applied finance models?
Thank you very much
As dumb as this sounds, learn how to learn. Barring that:
Dont learn a language, learn how to code - During school, spend time working on at least one big project, ideally working with different languages. It should be data intensive as well. The best way to learn how to write good good that is easily readable / scaleable is to do it in a sub optimal way, then go through the painful process of re-doing everything a bit better to accomodate some new feature, etc.
Take time to understand the finance side of things - I think there is this notion that quants don't have to be financially literate / can have a "its all just numbers" attitude - this is wrong. Take some corporate finance classes, do a few stock pitches - if you want to be a good quant, you need to have some intuition as to what make one asset look better/worse to buyers than another. Flows drive prices, and if you're an alpha quant your job will always be some version of finding out what drives flows.
Understand many things relatively well, and one thing REALLY well - Its a huge help to have one item in the toolbox that when asked "Do you know about X", you can barrel through any questions they ask trying to vibe check your knowledge. This doesn't have to be super broad - you just have to show some depth in one subject.
A good way to figure out what you need to know is to try and build a strategy that makes money. Make an attempt there and you'll figure out some basics of whats important / you need to get off the ground with this task very quickly. On a related note, people seem to always associate stochastic calc with quant finance - while I understand it likely gets used, for what its worth I have never met / know of anyone who actually uses this in any way to really make money.
Lastly, take some classes far outside your comfort zone (non-STEM / Finance) to broaden your horizons. College is to build yourself up as a person and grow along with the education it provides; It isn't just a stepping stone to get a job (if thats how you treat it, you're better off in the trades - its cheaper and faster). Meet cool people, have some fun, and occasionally drink too much - it doesn't last forever so take advantage of it. I've never heard anyone say "Man, I wish I slept more in college - I spent too much time studying any partying"
This is why alphas are crowded. Thanks for making my life harder
Thank you
Do you ever see people coming from risk roles? my role is heavily on the quant side but it does sit within risk, so not sure if that would be an instant turn off for many teams. I assume it would depend on how much I can show them that I've worked on the right skills outside of my job?
you can't make an internal move?
but yes, I have seen risk quants make the move
Anecdotally I have heard it happen, but often accompanied by a good bit of difficulty. From what i've heard, most risk work is very different than alpha generating work. Personally I have never seen the move, but again I know that it does happen on occasion. Sorry I cont provide more advice here.
Thank you for all the great info and detail. I'm currently recruiting a bit for MM's in addition to prop and had a question about your process in finding companies to apply to. Did you have a master list that you referenced or just dug around online to find smaller hedge funds? Obviously pretty easy to pull up the bigger/well-known funds, but for someone with a similar mindset finding the smaller places to apply or reach out to has been tough.
The best way to become aware of a lot of funds is just to follow the industry / read a lot. For example, the thread I posted above on nuclearphynance had the names of tons of funds that I'd imagine 99% of this forum has never seen or heard of. The only way you stumble upon these communities is by going down rabbit holes while reading about some other fund / strategy / etc that leads you there.
In terms of my actual "methodology" I had a spreadsheet that I would add names to whenever I saw a new fund that I had never heard of. Even just reading news articles can lead you to areas you didn't know existed. If you are really struggling to find names, heres a few ways to come up with names:
When in doubt, get people on the phone and be humble and enthusiastic - if you actually listen to people, they bleed very interesting information all the time; you just have to get them talking.
Thanks a lot for this, I find it very helpful for the junior audience on this board (including myself). A few followup questions:
- Do you have any advice on what to look for in the data as a quant?
- What is the best way to set up your process for researching new strategies/alpha? Do you feel MM's provide the necessary tools or u have to build your own thing?
- My understanding is that you have to look for new strategies all the time, since stuff stop working after a while. This seems like a cat and mouse game. Are there really strategies that work long term? If so, are they mostly based on fancy "alternative data" stuff or the more standard data can work too? If so, what percent of successful strategies are based on standard data as opposed to alternative data (understand the last part may be vague but roughly speaking).
These questions are generally strategy / process dependent - a few notes:
Wow. Someone who claims that there are *many* strategies that work all the time, they must be a cash cow. Jokes aside I think it's a very bold claim, and it wouldn't come out of someone with actual experience.
Dude anyone within the industry knows most of the useful datasets anyway- I don't think there is much of an edge in that anymore. More datasets you use, the better it is but there is no one "secret dataset" once can count as their secret sauce
Does this mean you are using primarily machine learning systems?
Yes, they are a very important part of our process.
What do you think about the ML "hype" these days? For example, Two Sigma hired Mike Schuster to head their deep learning team.
What fraction of it is something other than linear regression/PCA
Undercover meme - please PM me. I reached my limit for today for daily PM.
Question: while brainteasers and mental math are common interview subjects, they can be easily drilled. However, ML is a massive topic to prepare and so how much knowledge is needed for interviews (excluding OLS)? It seems to be that ML is used by many firms but rarely do I hear people getting grilled on it.
This is a great question. No one is going to expect a fresh hire to be an expect on a wide array of ML topics (barring some very specific seats, like ~10 a year maybe). Instead, I think there is one thing you need to have down 100% for any ML interview:
What you claim to know: If in your resume (or talking) you claim some knowledge (ie say your an expert on CNNs, or some ML method), then you have given me license to ask any question I want on this topic, and you should be able to get it right (which of course I can find a question where you wont - this is far game since you claimed to know it). They key around this is specificity - saying "Im an expect in NNs" is a whole lot easier to rip you on than "I have experience applying this model structure to datasets in this domain." The latter is much better as it tells me a) You have experience actually working with something, not just talk / fluff / you read ESL b) sets you up to answer questions specific / niche enough to what you just describe s.t. they should be layups for you, since you likely know more about the project than I c) the specificity tells me you are aware of what you don't know.
I'm not going to say that the quant space is immune to bullshitting, but being able to say "I don't know / I'm not familiar with that" to a crowd of smart people (or in this case interviewers) tells me that you're not going to lie to me on this kinda of matter if I hire you, which is valuable in itself. You still need to know some stuff thats basic (OLD< bias variance, main classes of models / what they are), but any of the "heavy" knowledge should be specific, and demonstrate that you can dive into one thing deeply. Remember, we want researchers who can do just that over and over again.
Hope that helps.
Very thorough and interesting post! On most internship requirements I see sth like "studying maths, physics, cs or another highly quantitative field". Does Econ fall under this? Would I have to go for pure maths or CS to become a quant?
ECON does not fall under this - there is considerable variance in the rigor of ECON programs from school to school (ie U Chicago vs. some softer LAC). AT the PhD level it is more respected (PDT seems to have an affinity for these guys) as these are more quantitatively focused.
For my personal option, I see ECON falling even more out of favor as many quant shops keep pushing out the old guard (closed form / linear model / etc) and fill up with more of the new guard (ML heavy / black box). I dont know how well this statement holds up industry wide, but this is what I've seen. Sell side may be more econ friendly due to an emphasis on "explainability" for regulators sake, but for systematic equities buy side I've seldom seen a solo econ major.
Thank you for the information! I'm currently completing a MSc in Physics, finishing in December, and I had a few follow up questions about becoming a quant. I've started reaching out and networking, but is there anything else I can do in the next 4 months that can help me prepare for recruitment? I was thinking of maybe working on a coding project of a basic trading algo, but I'm not sure it would be the best use of my time instead of revising math courses or learning about pricing models. Second, how does the quant trading industry look in Canada, or should I really be focusing on the US job market?
For QR roles, definitely get familiar with both exploratory data analysis and neural nets. Firms are rapidly shifting from hiring experts in math and physics who can whip out a proof of Black Scholes in an interview to experts in data science and ML who can make some neural network do the modeling for them in a nicely formatted jupyter notebook.
I think it depends on the shop. If you are looking at smaller teams where you will see the full picture than writing your own coding algo / being deeply critical of it and speaking about it in interviews, then that is a great show of interest far beyond what most people will do - this exact thing is what got my foot in the door. If you want to go to a place where you will be more of a "cog" (most new grads at de shaw / 2sig) which there is no shame in, then I would say crank it on the leetcode / questions from the books. The less structured the hiring, the less interview guide / leetcode there is.
As far as job markets go, I dont know much about canada which likely means there isn't much going on there - could be wrong but my understanding is that it goes US >>>> UK >> Amsterdam > France. Lastly, in reference to the person below this could not be farther from the truth in most places - I'd walk out of any place that wanted me to prove black-scholes.
I'll be straight up - How's the pay?
I'm deciding between IB or Master's in financial engineering -> quant as a college senior right now.
Thanks :)
If you are deciding between a MFE and an IB offer then that tells me you dont really care about either. Do IB.
Save
Removed for anonymity. Removed for anonymity.
Aperiam rerum est accusamus ut in. Dolores officiis asperiores sapiente aut. Eos corporis sed fugit placeat veniam distinctio eius.
Nihil magni sint sunt fugit voluptates velit quisquam. Eligendi aut laudantium eligendi quidem hic.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Iste autem fugiat quibusdam voluptatibus voluptatibus commodi voluptatum magnam. Odio nihil pariatur et dolor. Id totam quia assumenda eaque quidem.
Autem sunt dolor aspernatur officiis dolores facere nostrum facilis. Eveniet quibusdam dolorum reiciendis. Consectetur ea excepturi doloribus harum et. Magni voluptates qui tenetur non veritatis ut cupiditate numquam. Tempora est voluptatem in quaerat aut repudiandae vel quasi.
Est est sed nihil quisquam ipsum quo. Quibusdam voluptas facilis error hic illo fugit. Aut vel aut incidunt reprehenderit qui aut reiciendis. Dicta aut at dicta in id iste ab. Quo quia facere alias pariatur. Autem aliquid aut et aperiam.