What the interns have wrought, 2021 edition
blog.janestreet.comWhenever I hear about Jane Street and other such trading companies I wonder what do they bring of value into the world. It seems they're basically just working on increasing economic equalities.
Are you interested in a genuine discussion on this topic or have you already made up your mind?
For those genuinely interested, for the most part trading firms employing quantitative and automated strategies look to keep the price of derivative or correlated assets in line with their underlying securities.
For example a stock ABC might either directly or indirectly represent two other underlying stocks FOO1 and FOO2 and a trading firm will create a model to determine the relationship between ABC, FOO1 and FOO2 so that ABC = f(FOO1, FOO2).
The role of trading firms then is to compete both on coming up with an accurate model (one firm might believe that the relationship is really ABC = f(FOO1, FOO2) while another firm models the relationship as ABC = g(FOO1, FOO2) and yet a third firm models the relationship as ABC = h(FOO2, FOO3) for some third asset FOO3) as well as premium (a firm will trade ABC = f(FOO1, FOO2) + small_fee and pocket that fee), as well as engineering aspects such as identifying cheaper costs to trade, providing fast execution etc etc...
In the absence of efficient trading strategies or market makers whose job is to maintain this price, what you get is ABC will trade at a huge spread and trades will cause the price of ABC to fluctuate significantly. Both of these properties are vaguely lumped together and called "liquidity" so that trading firms are said to provide liquidity to the market. They ensure that ABC is available to buy and sell at all times at a price that reflects its actual value at every moment in time.
If there's some kind of economic inequality created by keeping the price of assets tightly coupled to one another so that their price reflects their actual underlying value then you are welcome to take some time to present that argument.
From my point of view having done this now for close to 15 years, I find it to be a very interesting and challenging career that lets me consolidate knowledge from a variety of different fields. I get to study and put to use knowledge of many areas of computer science from machine learning, graphics/data visualization, compilers, databases, computer architecture, networking, as well as other fields such as physics, economics, law. I feel very privileged to be able to work in a multidisciplinary field and be able to see almost in real time the effect of my work. I worked as a software developer at Microsoft and Google prior to this and while those were both comfortable and pleasant places to work, it didn't compare in terms of feeling like you had an actual impact on anything. I felt at both of those companies like I was being paid a lot of money to do basically menial jobs.
There is nothing menial about working at a trading firm.
You hit on so many great points in this post. Another important function trading firms (especially market makers like Jane Street) is to serve as ready-and-willing counterparties for the pension funds, endowments, and retirement funds of the world.
All of those firefighters, police officers, and teachers can't get their monthly check unless the pension funds converts part of their investment into cash, and the amount traded gets so large you need well-funded financial participants who will pay the pension fund a large lump of cash and assume the risk of unwinding the position.
The same thing goes for portfolio rebalances. Some smart and prudent retirement fund has run their risk model and decided that their sector exposures are off by like a few percent. They've got 2% too much financials, 2% too little tech, they want to move 1% from consumer staples to consumer discretionary, etc. We're talking about billions of dollars being moved right here! They want to use the money from the assets they sold to purchase the assets they bought, but they want the transaction to happen all at once... not in two parts. Again, this is where the quant funds and market makers of the world come in.
It's easy to think that finance is all fat cats, hedge funds and multibillionaires, but anybody with a retirement plan, a pension, ETF holdings, or mutual fund holdings, benefits from the work these quant funds do.
Just because financial firms provide legitimately important services, doesn't mean the value they capture is commensurate with that value creation.
So yes, it's a leap of abstraction for most people to grasp that 'market making' etc. is an important function, and that there's legit value created.
... but the amount of surplus capture is gigantic and that's the problem.
Fouquet became Louis IX 'CFO' and somehow at the same time amassed massive wealth possibly illegitimately and so he lost his head very quickly.
It's always the people closest to the piles of Gold that end up with all of the Gold. There are a million ways they justify it, and in free markets where information and relations are 'the power', well, they have it all.
Usually it's 'generally very talented' people involved, of that I have no doubt, but that again doesn't justify necessarily the surplus capture.
> It's always the people closest to the piles of Gold that end up with all of the Gold.
Honestly it's a lesson I've taken in my career so far - those who are closest to the money flows get to siphon off the most for themselves. I'm by no means that close, but it certainly informs what I put myself forward for.
Does the difference in skill, effort and output justify it? No, and many of the people working close to the money are no better (or are often far worse) than people further away, being paid less. But people are willing to pay them more, because they are also close to the larger flows of money.
If you count the trading firms that don't do so well, are the people in finance really making off with all that much money, or is it the best firms taking all the sunlight from the middle and worst firms?
Sure, but then you'd have to count the regular businesses that don't post heavy profits of which there are many.
Also either it's 'market making' or 'trading' I think it should never be the same thing under one roof and that market making should have a fair bit of regulation and transparency.
Are you aware of market making actually means?
Because your statement about "either it's 'market making' or 'trading' I think it should never be the same thing under one roof" is like saying "it either gotta be 'playing soccer' or 'kicking the ball', they should not be the same".
Sure, "kicking the ball" is done by many different sports and in many different ways, but it is literally an integral part of playing soccer, so you cannot say "either or" here. The same.
The whole point of market making is trading securities at both ask prices and sell prices, thus providing liquidity to the markets. So they essentially are making profit off bid-ask spread, but doing so reduces bid-ask spread and decreases market inefficiency (thus converging on a more accurate value of the security in question).
And to your point about regulation of market makers, they already have more than just a "fair bit of regulation and transparency".
Yes, of course, it was a poor choice of words on my part, of course market making implies 'trading', I meant to imply advising parties on both sides of a trade, or other forms of conflict of interest.
Look into info about Jane Street and other HTF firms' manipulation of OTC stock prices via dark pools, as well as abuse of the authorized participant status to use the ETF mechanism to naked short sell (not just index arbitrage). The liquidity that's being provided is at the expense of real price discovery and is being used to con retail investors.
If you search for "David Lauer interview", he's a former Citadel HFT that has spoken out about how the funds operating right now are a world away from the platonic ideal of market making you speak to.
I've been in the field for a nearly two decades now, and it's highly recommendable. There's a game-like element that appeals to a certain type of person.
As for doing good in the world, there's an awful lot of businesses that really only contribute via tax. They do something where the entire benefit, sometimes more, is captured between the business and its customers.
But to say it specifically, you need market makers, because if they weren't there you'd have to wait a long time for someone with the opposite intention to you to show up, and that has a huge cost. And keep in mind MMs are only one part of the capital apparatus that we need in the world. This isn't just fluff, I literally had a friend phone me last week looking for money for his African startup that has had to turn down business for lack of capital.
The how of it is a long story as well, but quite an interesting one that spans a number of fields. Coding, economics, finance, and so on.
Another practitioner, my 2c on high frequency trading:
1. Market making (providing two sided liquidity in a market), and in general providing liquidity (even one sided) is a real benefit that (some) trading firms provide (e.g. Virtu, but also many proprietary shops).
The market making function solves the time-mismatch between client A who wants to buy now, and client B who would want to sell in 1 hour.
The liquidity provision is what allows, through pay-for-order-flow, retail investors to get the best current price for AAPL through their Schwab app. Not only that, they do not have to worry about the order execution (how long would it take to get filled), and their fees are low or 0 (because someone is paying Schwab for the order flow).
2. A similar, but slightly different trade is what ensures that ETF are priced "correctly". This is also a boon for retail investors.
In order to ensure the ETF price are in line with the price implied by their underlying assets, someone need to do the "index arbitrage" trade. Again, retail gets a nice useful product (e.g. SPY which tracks S&P500) instead of having to buy a lot of stocks (naively, 500).
3. And then there are directional trades, which are basically trying to "arbitrage away" moves in correlated prices. Is there a value in this? Maybe, but it is just accelerating existing trading realities. Someone would have done it anyway.
I am guessing the perception that the industry adds no value is because of stories about the ridiculously low latency number - trade in the blink of the eye, reacting in nanoseconds, etc.
And I tend to agree.
If the exchanges changed the way they matched orders (e.g. instead of First In First Out/time priority ["who came first"]) to a procedure that introduces randomness that eliminated the advantage from nanosecond/microsecond differences, it would make no difference to the world.
The prices would still be "good enough" (no retails participants care if the prices settled within 100 nanosecond, 1microsecond, or 1 millisecond).
So to sum up, does it bring value? Yes, but it does not mean there is no silliness involved.
I personally think it's refreshingly honest to focus on moving a number up and to the right, rather than pretending to "make the world a better place" while the bottom-line is to move a number up and to the right.
Exactly. Trading has a clarity of purpose that I just love: unlike other jobs no one is lying to themselves, no one is claiming to help other people. We know there is only one reason why we work together: to make as much for ourselves and the firm as humanly possible.
Look at how much trading costs have reduced for all investors over the last two decades due to reduce spreads due to automated market making.
they improve ocaml libraries?
In a capitalist economy everyone is working on increasing economic inequality. It seems for simplicity sake the question of what sorts of industry generate value should be separated from a question of whether corporations should be contributing to inequality.
According to wikipedia they have only 1200 employees.
And 87 interns, that's quite a lot.
Given the level of granularity in that code, I can't imagine what they are all doing.
How do you have 87 people tweak a code-base and then just leave?
There's an incredible amount of sophistication in some of those project, surely they're not all used, but imagine the team you need just to maintain any of that. With only 1200 staffers, how many are in Development?
I think there must be a huge amount of surplus spend here. Basically, they are swimming in cash, hire the best interns, get a few nice things but mostly it's not that important. But if those marginal contributions eventually do provide a tiny bit of value the amount of leverage involved is so huge that it can make them a fortune.
I think at most top of line places, internships are viewed as a two way interview. Firms get to see how the intern performs over 8-12 weeks, and the intern gets to see company culture/fit for 8-12 weeks. The work the intern is doing isn't the goal of the internship.
I understand that, but $45K is a huge amount of money to spend on that, and especially for so many.
I worked at a blue chip we hired a lot of interns and we put them to work, man they ran a lot of things.
That said, it was hit or miss.
Well, they're also effectively spending for an extended job interview.
Do the interns get paid?
> Like most companies, Jane Street also doesn't discuss pay. However, as we reported last year, pay for U.S. trading interns at Jane Street in 2020 was $14.5k a month, plus an extra $500 a week work from home stipend (making $16.5k as a monthly total) for.
From https://www.efinancialcareers.com/news/finance/jane-street-p...
They’re paying their interns how much? Either finance pays way more than I thought, or that’s wildly off - for comparison as the most senior developer in our London based organisation, having been here since the beginning, I make about $11k a month.
Former Jane Streeter: yes, finance can pay very well. Of course there’s a lot of variance, but at least when I was there Jane Street made an effort to be the very highest in terms of compensation for interns. The numbers don’t seem off to me. Yes, I understand how ridiculous that is
I don't think it is ridiculous if you can afford to and want to pay the absolute top line to get the "best" (best by their metrics not everyone else's - not trying to spark a war).
If you are ultra profitable then the difference between paying market rate and top of market rate might be double the salary and that would still be a rounding error on a spreadsheet somewhere.
I guess it also depends how sensitive your output is to the quality of the developers, lots of enterprises I've worked for you'd be hard pressed to tell the difference between a competent developer and an excellent developer since the end result is from a distance pretty much the same in that case it would make no sense to pay more than will get you a bunch of competent developers.
My impression from the outside is that janestreet does see that value.
Some what correlated I love it when companies say "We hire the absolute best developers" and you ask what they pay and market-rate is the answer.
JaneStreet remains fascinating to me though because while they do programming and I'm a programmer we are almost different species because of the problems we solve - which makes it interesting to me.
what's ridiculous about it?
As a senior developer in Europe, I'm considering becoming an intern.
Definitely - although the salaries here (at least in Berlin) are raising from what I can see as well
Jane Street interns are known (maybe not in this conversation :P) to get paid exceptionally well. A typical software engineering intern at a large Bay Area company is probably making something approaching $10k.
That’s about the ballpark for a paid SE intern in Germany (per annum, of course).
10k per month in the Bay Area.
Yeah we’re a small startup and we pay $120k / 12 per month for interns. Otherwise it’s difficult to get interns at the level we need. Really not surprised finance pays more.
180k annualized salary for the top few students from MIT, Waterloo, Stanford etc is a steal. They might lack sophistication in navigating bureaucratic processes but I guarantee you they are far more technically competent than even your average senior engineer at a FAANG company. Most companies don't really know how to utilize people with an elite level of skill but Jane Street clearly does and is willing to pay up for it.
Jane Street can’t pay any stock (proprietary firm) so they have to pay a lot more cash.
Do UK/European organizations typically give employees stock?
14.5k is almost the monthly base pay for a senior software engineer at a FANG in Bay Area/NY/Seattle/ZRH (not counting bonus and stock though - which make a significant difference).
So it does seem very high, but not impossible?
I believe Salesforce offered interns in SF something like $8600+/mo and an option of company-provided housing or an $8k stipend, if they took the latter then their effective pay would be a bit over $11k/mo. So yeah, it doesn't sound unreasonable. Internships only lasted 3 months, so it's not a big burden on the company, plus it gives the intern a salary proxy for what they can expect as minimum if they're invited back for full time the next summer after graduation.
No way, I get more than that at an entry position outside of FANG. Senior software engineer pay starts at $25k/month and goes up from there.
Edit: I just noticed base pay–then yes, that's about what it'd be, but base pay tends to cap out very quickly so it's not a good measure of compensation.
I moved to the bay area at the same time as taking a senior engineering role for an Eastern European startup on $100k - feel like I missed a trick lol :P
To be fair I don't care about salaries as much as half of HN appear to, I'm quite comfortable on that with a wife on a postdocs wage too.
As an American who worked in London and at a London company, London pay is always crazy low compared to what I’d expect.
I think you're not up to date on the engineering market in London. Google London pays $11k/m (£100k/y) TC to strong new grads these days, and it's not uncommon for exceptional ICs in finance to earn up to £1M/y
Jane Street candidates are already top shelf candidates, and can pretty much write their own tickets - so it's in the best interest of companies like JS to capture them for potential full-time employment, before they graduate.
A solid salary helps.
Jane Street pays a lot - I know of people who earned half a million a year when they worked there in their early 20s
Jane Street pay graduates in London several hundred thousand pounds.
> the most senior developer in our London based organisation, having been here since the beginning, I make about $11k a month.
Pre or post tax?
As a London fintech contractor (until recently) I have able to pull in around $22k USD pcm pre tax, or very, very roughly $14k USD pcm post tax - contractors have various ways to manage tax load and time off so it's never a direct translation.
$16.5k pcm for an intern still seems utterly crazy, given my starting annual salary of £21k 20 years ago, but if there is a) a ton of money flowing through the company and b) the intern is able to help them bring in more than that over the course of a year then... I guess it's worth it to them, and that's all that really matters.
Finance pays at least as well as FAANGs (anecdote: in 2017 I had an offer from a FAANG and a finance company, the finance offer was slightly higher, and the FAANG flat-out refused to negotiate a finance offer), and also my sense is that salaries outside the US aren't keeping up at all with the US.
The salary numbers on https://levels.fyi seem to be plausible enough for US FAANG salaries, if you want an idea of what they look like.
I’m far from the most senior developer at my company and I make far more than that here in the U.S. To the point I would be insulted if I got a job offer for $11k a month. The thing is, salaries for engineering talent is just really low in London.
I saw a ad for a job at an HFT the other day claiming it paid $850k a year, not sure if that was base of TC though.
From what I hear though, interviews are hell, worse than FAANG or any tech companies and probably extremely competitive.
The lowest pay recorded for Google in London is $124,000, for L3, new grad. The lowest for L4, which you would surely get as an experienced hire, is $155,000. You get paid roughly as much as the lowest paid permanent fulltime staff at Google.
https://www.levels.fyi/comp.html?track=Software%20Engineer&s...
A couple of other notes:
Finance non-competes seem to be MUCH more restrictive than tech. Also, you have to average 5-year comps... because Finance pay can be very volatile. Also, in tech, many more companies try to filter out non-team players during interviews.
It sounds about right — finance engineering interns from our school like 5-6 years ago after our 2nd/3rd years fielded offers for $12k-$14k a month, with a housing stipend included
Of course. How would they attract quality people if they didn’t pay anything?
Good mission statement and socially respectable goals? It doesn't have to be about money.
Even people who work for charities need to eat.
Talk is cheap, but it takes money to buy a farm.
I wonder how’s the work life balance there especially if you are full time employee and what’s it like to work their as a Linux engineer.
Is there anything similar to DataFetcher in python? I have been considering using Prefect which is a workflow system with caching.
Found something if anyone else is interested: https://github.com/syrusakbary/aiodataloader
Where are we standing with multicore Caml?
Closer than ever :) There is a monthly post on ocaml's discourse that reports progress in upstreaming patches into mainline ocaml (see https://discuss.ocaml.org/t/multicore-ocaml-july-2021/8232 for the latest). I think it'll happen next year.