Forums  > Careers  > starting as a Quant Researcher buy side a smart move? Dying industry?  
     
Page 1 of 1
Display using:  

candide


Total Posts: 4
Joined: Oct 2010
 
Posted: 2020-10-31 15:41
Dear all,

I'm finishing a postdoc (physics/ML) and have been offered a tenure track Assistant Prof position in a (top20) university. I have also been offered a QR position [alpha generation / medium frequency] in a reputed fund. I think that both careers would roughly be as intellectually stimulating:

1. much more freedom/flexibility/stability in academia [and a probability ~80% of getting tenure in that university] but probably a slower pace (a bit of publish or perish nonsense), quite a bit of teaching [that I do not particularly enjoy] and the usual grant writing and fair load of admin etc.. That's very exciting to run a small lab, but this also probably means doing less actual research/coding in a few years time [more expectation in terms of obtaining grants / supervising students / etc..]. I do really enjoy coding / solving hard pb / etc..

2. salary as an AP is not bad at all (~150K) and would definitely be OK for taking care of a small family. Starting package in the fund is about 2 times that [and likely to increase more rapidly than in academia]. There is also a long non-compete, which worries me a bit.

I have also been talking to a few AI-labs on the IB side, and also to a few tech companies. Salary a bit lesser than QR buy-side, and I found that what they were doing was a bit less interesting to me [a lot of it is fitting large neural nets / etc..]. But definitely an option I am keeping in mind.

Now, my main questions are:

A. is being a QR in a fund is really a smart moves nowadays. The pay is fantastic now, but how long is this industry likely to last really?

B. If I start as a QR, I do not think I have the ambition to become a PM. I am seeing a QR position as a "well paid engineer" and I quite like that. But what's the typical trajectory? Is doing that during 30 years common?


Thanks!

chiral3
Founding Member

Total Posts: 5180
Joined: Mar 2004
 
Posted: 2020-11-01 15:05
It could maybe be called the "ten year rule (of thumb)". Gist is that every ten years of your career you look back and say "I could have never predicted (all these changes)".

Don't get too hung up on trajectory; although I will say that where you begin greatly influences the sort of things you wind up doing. Thing is, things are so weird these days I think it's hard for anyone (starting out) to have a good sense of what a good track is to be on. Even your academic choice is overshadowed by the bullshit happening with universities these days. Buy-side is definitely better than the the sell-side, if only for the fact that fact that over-regulation has choked the sell-side to death and most of the people left there, assuming they aren't doing something regulatory, are clipping a coupon. If it's money you're after, private equity five plus years ago would have been smart. PE has been benefiting from regulation but it's been competed away for anyone at your level.

Answer to A: you'll get experience, tools, learn if the culture is even right. You'll know nothing about how a business works after a couple of years. There will always be QR jobs, no matter what form they take.

"..." B: Yes. You run into the "pay problem". Most of the people that I know that have many years in QR and its ilk are highly paid up so some murky cap. At some point they want more and things start to go south. By more I also mean creative control. Then the lack of understanding in all the dimensions start to become uber apparent and things go to shit. At this point they go back to teaching, or writing books, or doing adjunct work, creating certification courses, lecturing at conferences, ... all if this worked out well for a few people you've heard of from maybe 2005-2012ish? The irony is that these people, if they ever had jobs at all, were low-level and / or crappy at their jobs, but they created these little personas. Neither path sounds fun to me.

Don't think too far ahead. Try to apply simple rules for your decisions:
1/ is this a place I'll be happy?,
2/ do the people and the work inspire and excite me?,
3/ is the work deployed and used, not just shelved as research?
4/ do I have the autonomy, agency, and tools to succeed or fail on my own?

Nonius is Satoshi Nakamoto. 物の哀れ

Maggette


Total Posts: 1269
Joined: Jun 2007
 
Posted: 2020-11-01 18:56
Echo chiral on the 10 years rule. Don't waste too many CPU cycles on decisions that involve long term predictions. The decision is also not forever.

My personal choice woul be the qr job. Especially if you like to focus on solving problems. But I never worked in academia. I am just put off by some horror stories who worked there or do still work there. But also know some people who are really happy there.

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

EspressoLover


Total Posts: 461
Joined: Jan 2015
 
Posted: 2020-11-02 16:15
> I have also been talking to a few AI-labs on the IB side, and also to a few tech companies. Salary a bit lesser than QR buy-side

Have you looked at the FAANG companies and the major unicorns. I'd be pretty surprised if they don't beat the finance offers. An ML PhD with coding chops can start as an E5 at Facebook, which comps around $370k. If you're good, and get in a high-impact area like ad targeting, you should be able to hit E7 in four or five years, which comps around a million.

And that's not even taking into account being roguish about it, and flipping jobs every year or two to negotiate better packages. Forget who it was, but one of the senior executives at Microsoft said they're paying top-talent deep learning candidates more than first-round NFL draft picks.

Overall FANG-tech is probably less intellectually rewarding than quant trading though. Depends how much you like the pure life of the mind, versus "getting shit done". Tech work is a lot more emphasis on "doing" rather than "thinking".

Good questions outrank easy answers. -Paul Samuelson

sharpe_machine


Total Posts: 69
Joined: Feb 2018
 
Posted: 2020-11-02 16:26
> you should be able to hit E7 in four or five years

Betting on hitting an E7-equivalent in <= 5 years is ... questionable to say the least. Especially in the G.

gmetric_Flow


Total Posts: 43
Joined: Oct 2016
 
Posted: 2020-11-02 17:02
The current regulatory environment creates this scheme in tech, but if history is any indication, the compensation will be competed away rather rapidly (to chiral3's point, it's not much different than the PE scenario from a few years back). It's difficult to bet against big tech, but I'm skeptical, especially given the compensation–talent ratio.

nikol


Total Posts: 1235
Joined: Jun 2005
 
Posted: 2020-11-02 18:03
Google is a bank now (or whoever is)
FB just narrowly escaped the fate of FED

Others will follow. So, very likely FANG will follow the same path as HFs. Basically they do same MM strategies actually without realising they do.

Jurassic


Total Posts: 379
Joined: Mar 2018
 
Posted: 2020-11-03 19:04
@gmetric_flow what are the forces which are going to compete it away? Im not sure its that easy, especially at Google, to pass their coding questions, which surely restrains the amount of talent? Why would be being a quant be any different and the comp there is still high?

hftpm


Total Posts: 6
Joined: Oct 2013
 
Posted: 2020-11-03 22:23
HFs are not dying, but current AI mania is going to end soon after they realize for the nth time that this shit don’t work for alpha generation. If you cannot come up with signals using whatever math you will be fired. For somebody who doesn’t have market experience it’s tough.

If you want to apply AI in industry go to big tech. Otherwise, go to academia.

It’s not like you want to make it big anyway from the look of things




Maggette


Total Posts: 1269
Joined: Jun 2007
 
Posted: 2020-11-03 23:29
Well. You could say that about almost any newcomer to the industry.

Alpha is very very hard to find. At least from the small sample I have experienced I would infer that without adding stupid risks (like being short gamma all day) most HFs don't add any value to a portfolio that broadly diversified regarding its factor exposure (AKA beta).

Everything that is hugely profitable seems to be MM in nature and is part of closed funds or prop shops (IMHO there is where the true market intelligence is located).

The man sounds like a well trained scientist with ML and Stats skills. I don't see how he should be at an disadvantage to anybody else. MAybe even to guys with business experience.

So far I wasn't overly impressed by many people with a prominent career in the IB and HF world (Prop Trader at Big Shite Bank, then Head of Equity Derivative at Bullshit Bingo Bank, and then founding Partner and Head of Quantitative Strategie at Same Old Stuff LLC...you know these guys). Small sample size though and of course some exceptions to the rule.

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

gmetric_Flow


Total Posts: 43
Joined: Oct 2016
 
Posted: 2020-11-04 00:17
@Jurassic If you assume that the tech companies will continuing growing at similar rates and their demand for talent scales accordingly, I guess it won't be - although that doesn't seem too reasonable. Certainly not saying it's easy, but my perception is that the average technical competency level per dollar of compensation is much higher in quant finance. Would love to hear if you (or others) feel otherwise, my sample-size isn't particularly large.

candide


Total Posts: 4
Joined: Oct 2010
 
Posted: 2020-11-04 02:21
Thanks a lot for these answers.

What worries me a bit about the QR position is that I have this feeling that only very few new ideas have been successfully exploited over the past 10 years, and it is unlikely that all these recent non-linear methods [neural net / etc..] have much to bring -- and it is just that in that type of regime (mid/high frequency), no fancy method will do much better than what exists. For example, it is not clear that all these recent important advances in NLP could be at all exploited successfully. When I see known and respected (?) quants such as Marcos Lopez de Prado [former head of ML at AQR!] talking nonsense about "Quantum Finance", or stating on twitter than logistic regression is dead (I am paraphrasing), it all sounds desperate (hence my wondering whether the industry is dying).


While talking to a few funds, I had this impression that they still hoped to come-up with relatively new methodologies [ie. not necessarily better, but at least not highly correlated with their existing set of methods] -- and it is worrying to me because I am not sure I would be able to meaningfully contribute, should I join such a firm --> I will get rid of in a few years time, with a 18/24 months non-compete.

Maggette


Total Posts: 1269
Joined: Jun 2007
 
Posted: 2020-11-04 11:39
I am not exactly confident that Marcos Lopez de Prado is really respected. I liked some of his ideas on Tree Clustering in portfolio allocation and some trivial things he says about CV and back tests.

But other than that I wouldn't think he is an indicator for what professionals in the field do.

As an disclaimer I should add I moved away from finance/trading 6 years ago (living in Germany and not willing to relocate..there is close to nothing interesting going on in Finance here).

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

chiral3
Founding Member

Total Posts: 5180
Joined: Mar 2004
 
Posted: 2020-11-04 14:04
I'd like to avoid mentioning names, but since you brought it up... that's an example of exactly was I was talking about. AQR is a good example of something many well known managers did back when times were good. They'd keep these people around that didn't really have any real jobs, but they hit the circuit and write books. It was this weird affinity many of them had with having this personality running around with the convex and deeply OTM payoff profile that they'd get picked up in the popular press for some wonky idea. I can think of 5 to 10 people attached to large and well-known funds in the past ten years that fit that profile and, for someone viewing outside/in, they are probably the face of the business. Some of them are really good people that I consider friends, some I could do without, but I couldn't tell you what any of them actually did for a living. They all clipped a modest coupon about equal to the option on that book, paper, or article.

Nonius is Satoshi Nakamoto. 物の哀れ

sharpe_machine


Total Posts: 69
Joined: Feb 2018
 
Posted: 2020-11-04 14:46
> it is not clear that all these recent important advances in NLP could be at all exploited successfully

Personally know several groups who do use modern NLP to trade around earnings reports. The problem is that the PM collects the majority of upside, while his junior "rockstar" NLP-gurus earn "just" FAANG + risk premia (still good money for people who target to be a well-paid engineers).

gaj


Total Posts: 115
Joined: Apr 2018
 
Posted: 2020-11-04 14:48
> Personally know several groups who do use modern NLP to trade around earnings reports. The problem is that the PM collects the majority of upside, while his junior "rockstar" NLP-gurus earn "just" FAANG + risk premia (still good money for people who target to be a well-paid engineers).

what's stopping those rockstars from becoming PMs themselves?

sharpe_machine


Total Posts: 69
Joined: Feb 2018
 
Posted: 2020-11-04 15:35
> what's stopping those rockstars from becoming PMs themselves?

I'm not sure it is possible to answer this question. Those people do research in NLP, I'm not sure they are involved much into the rest of the trading pipeline. Assume you are a PM which earnings depend on a performance of that strategy. And you would like to improve one particular feature of this strategy. So, you would like to hire some expert in this feature, but do not really want him to become your competitor in the nearest future. So, one of the best personalities to hire to this role is a scientist who really wants to continue doing science but be paid much higher.


Moreover, not everyone wants to be a risk-taker which is a slightly different personality.


Jurassic


Total Posts: 379
Joined: Mar 2018
 
Posted: 2020-11-04 17:16
@gmetric_flow would also be interested in others views

@candide also if you are looking for a very academic, research style environment you may be very disappointed in most quant funds

EspressoLover


Total Posts: 461
Joined: Jan 2015
 
Posted: 2020-11-04 17:52
> what's stopping those rockstars from becoming PMs themselves?

Honestly, it's extremely stressful to be directly responsible for PnL. Even if your payout is ultimately tied to a volatile process, cashing the bonus checks every quarter is a whole lot less emotionally draining than watching the day-to-day time series. I think it's relatively rare that you find a single person with the personality confluence of a curious, thoughtful scholar with a cold-blooded calculated poker player.

Plus like @Sharpe mentioned, most of the major shops do a pretty good job at keeping people siloed just enough that they don't see the full picture.

Good questions outrank easy answers. -Paul Samuelson

hftpm


Total Posts: 6
Joined: Oct 2013
 
Posted: 2020-11-06 17:45
OK, obviously there's not much understanding what a typical fresh PhD level quant does.

This is Nick Patterson, former head of statistics at a certain best pension fund in the world. It's also the world's "best math and theoretical physics dept" and everybody works together. They even pretend not to hire people with Wall St experience.

In a typical multi manager HF quants and programmers are the PM's bitches and are not allowed even to talk to people from other groups.

If you think you gonna be doing some highfalutin science stuff at any fund, you are deeply mistaken.

"I joined a hedged fund, Renaissance Technologies, I'll make a comment about that. It's funny that I think the most important thing to do on data analysis is to do the simple things right. So, here's a kind of non-secret about what we did at renaissance: in my opinion, our most important statistical tool was simple regression with one target and one independent variable. It's the simplest statistical model you can imagine. Any reasonably smart high school student could do it. Now we have some of the smartest people around, working in our hedge fund, we have string theorists we recruited from Harvard, and they're doing simple regression. Is this stupid and pointless? Should we be hiring stupider people and paying them less? And the answer is no. And the reason is nobody tells you what the variables you should be regressing [are]. What's the target. Should you do a nonlinear transform before you regress? What's the source? Should you clean your data? Do you notice when your results are obviously rubbish? And so on. And the smarter you are the less likely you are to make a stupid mistake. And that's why I think you often need smart people who appear to be doing something technically very easy, but actually usually not so easy.

At my hedge fund, which was not a very big company, we had 7 PhDs just cleaning data and organizing the databases."

http://www.thetalkingmachines.com/episodes/ai-safety-and-legacy-bletchley-park







chiral3
Founding Member

Total Posts: 5180
Joined: Mar 2004
 
Posted: 2020-11-06 19:49
"...certain best pension fund in the world. It's also the world's "best math and theoretical physics dept and everybody works together. They even pretend not to hire people with Wall St experience."

Funny.

I'll disagree with some of what you said. When I started out as a fresh quant I was a data and model jockey. However, having close to two decades of experience hiring fresh PhDs across several industries (sell side and buy side) I'd say it depends on what you're doing and where you play. I've hired plenty of fresh PhDs that started immediately with heavy PDE, portfolio optimization, etc. Data monkey, PM lackey, backtest jockey, ..., absolutely, but by no means the rule. Certainly at your fmr employer I can understand why many PhDs just scrub data.

Nonius is Satoshi Nakamoto. 物の哀れ

rod


Total Posts: 430
Joined: Nov 2006
 
Posted: 2020-11-08 14:35
@candide:

In my humble opinion, like most people transitioning from academia to finance, you focus a bit too much on ideas. It's not that ideas do not matter, however. It's that far too many people in finance have absolutely no taste and believe that technical work is beneath them. As a consequence, firms can end up with a multitude of databases that produce different values when queried. Since many people are hopping to better-paying jobs every few years, whoever made key decisions did leave several years ago and documented nothing in order to increase the cost (to the firm) of having them fired.

Sure, linear regression is easy... in textbooks. In real life, often, people cannot even agree what the data are.
Previous Thread :: Next Thread 
Page 1 of 1