Comment by lordnacho | Hacker News Reader

Comment by lordnacho | Hacker News Reader

Parent Full thread

Quant dev here.

Note there's broadly two kinds of quants.

One is a derivatives quant, basically someone who spends a lot of time working out the values of derivatives. So you're looking at ito calculus and iterative methods to find the value of contracts with weird clauses (Knockouts, Cliquets, vanillas, etc), and then building spreadsheets to calculate the hedges for those trades as time goes by.

The other is strategy quants. Here you're looking for some way to beat the market. So you're reading in a bunch of data from various sources and applying some sort of quite open-ended analysis on it to come up with rules for a trading system. You then build a framework for trading such systems, which can get very involved eg execution algos that read realtime market data.

I started off in derivatives as a trader but quickly moved on to the other. They're both fairly deep, and both require you to know coding and math.

The questions here seem to apply mostly to derivs quants, but strategy quants have found many things applicable.

Serious question... do people really enjoy working on this stuff? Do most quants go to work thinking it is fun to work on mathematical models to optimize financial transaction. Or is it just a job that pays really well towards long term financial independence to do what they really want?

There's some interesting things about this work:

- You get to know something that isn't published. You find a strategy, you don't tell anyone. Sure, you can talk about it in broad strokes (we're trend following...) but it's pretty unlikely you'll ever tell anyone enough for them to be able to replicate it. Except people you trust, of course.

- You're always thinking about philosophy of science. How do I know that this series of steps I'm doing is not just a monkey and a dartboard? Again, there isn't going to be a specific written paper that tells you the answer.

- At one end of it, you are working with cutting edge technology. Anything that shaves a microsecond off the speed is useful, and you can spend a lot of time optimising such things. I've been staring at a C++ solution that is 100% in house. No STL.

- At the other end, you are working with data in very modern ways. All that ML stuff in the news is interesting to quants, because if there's one thing ML papers are about, it's how to avoid overfitting. Just a few years ago people in the field had not come across ML, or at least the point of view that ML brings. Plenty of other things had already come up, though, like information theory, time series econometrics, signal theory, and so on.

- About derivs quants, it's not my cup of tea, but I can see how others like it. You're basically pricing things hoping that you've noticed something the counterpart hasn't, or hoping you've found a cheaper way to hedge than your competitors, or simply hoping your salesman is a better salesman than the other guy's.

This is good stuff, thanks for articulating your experience in the role.

Honestly, this is a great pitch for your line of work. It sounds like you really enjoy it.

Yes it's an awesome job, and not just because of the money. I've worked for small shops that weren't very profitable and didn't really make a lot those years, but still loved the work.

I could see derivatives pricing being boring, not as much room for creativity there. The strategy/trading quant work is like a super open-ended science problem with a constant stream of new challenges as you try to outperform yourself or the competition.

FWIW, I don't think these are great questions for strategy quants. Some of the best people I've worked with a.) used very, very basic ML techniques b.) were just naturally smart/clever enough to engineer great features. and c.) had incredible attention to detail, didn't "hand wave" away the dirty work like checking/cleaning data or executing the model in practice.

I've heard the working hours are terrible with 12-14 hours workdays not at all uncommon. Is that true? Is the work pressure very high?

If you're making money everything is relaxed. If your strategy is losing money things will be very tense. That's just how it is when you have skin in the game.

>> 12-14 hours workdays not at all uncommon.

> That's just how it is when you have skin in the game.

MD here, and prior military. That's kind of how it is in any profession where you have skin in the game.

Also, for those interested in physics: my undergrad quantum mechanics professor stopped in the middle of lecture once and announced to all of us, the only section of physics majors (17 in all): "Odds are no more than 1 of you will get a PhD in physics. If you want to keep doing math, most of the jobs are on Wall Street and Vegas." And just turned back to the board and kept writing. Probably something about a Hamiltonian...

What's MD?

I assume doctor.

Can also mean 'managing director' in finance and other fields

> I've heard the working hours are terrible with 12-14 hours workdays not at all uncommon.

Derivatives quants have perfectly normal hours

About questions not being great, do they match what you could expect in a real world interview for a strategy quant position?

Everything in sections 1-3 looks like pretty standard entry-level strategy quant questions.

Not a quant, but:

Imagine playing a huge board game with a bunch of very smart (as in, MIT math PhD smart) players that spend every waking hour trying to come up with strategies, and every time you come up with a good strategy the entire metagame of all players quickly adapts to its existence so the game stays interesting (except for some that lose before they can adapt and have to leave the game.)

Factors in a good game strategy might include fast thinking, every historical move a player played in the game, human psychology, poker-type bluffing, reverse engineering algorithms based only on their outputs, correctly predicting the future of industries and companies, headlines on the news, sentiment analysis on social networks, predictions of election results (and stuff like the Brexit referendum), as well as the complicated rules of the game itself, and Sicilian Reasoning (http://webdocs.cs.ualberta.ca/~darse/rsb-results1.html).

Some things you might do as part of a strategy: charter a fast ship to send information to your friends in another market before it becomes public, so they can buy a lot of stuff before it's price goes up; buy a donut every week in the same shop and keep track of receipt serial numbers to estimate changes in a franchise's weekly sales before anyone else knows them; use satellite imagery of Walmart parking lots and some computer vision algorithms to get a very good estimate of their sales; borrow $10Bn from your friends and use it all to sell British pounds in a very short period of time, forcing the British government to stop pinning the pound-dollar exchange rate and making $1Bn in the process; install a spy network throughout Europe so you get information on the result of the battle of Waterloo a day before the King's intelligence service, spread a rumor that Napoleon won, buy everything that got cheaper because of the rumor, then the next day sell it when the truth that he lost arrives and use the money to more or less buy everything of value in Britain.

At each point in time there is a clear tracking of scores so you can know exactly how well you're doing relative to every other player.

And the best thing is, you get paid the best salaries in the market to play this game.

I love being an entrepreneur because of how open the rules of the game are and how many creative moves I can pull off to get advantage without having to convince some judge or committee or boss that this is a good idea. Strategic quants have all this and the additional benefits of immediate feedback and a great salary with low risk.

A good example of real-world application of the donut-receipts idea above is the German Tank Problem[0] from WW2.

[0] https://en.wikipedia.org/wiki/German_tank_problem

All the ideas I gave were things that happened (or at least, that I read on the internet from reliable sources.)

Chartering fast ships is from 1790, see e.g.: http://dewinforex.com/forex-basics/high-frequency-trading-fi...

Can't find the link for the donut investor, but I'm pretty sure it was somewhere in Matt Levine's (highly recommended) Money Stuff column.

Sat imagery of parking lots appears all the time in Matt Levine's column, there is a company selling this data to hedge funds today.

Selling pounds is of course George Soros, probably the most famous story in Finance.

Wikipedia claims the Rotschild story about the Napoleonic wars is fake. I read it on Quora here: https://www.quora.com/Whats-the-shrewdest-smartest-maneuver-...

I got most of the references, but the donut one just seems very interesting. Please tell me if anyone here manages to find it. I'll be looking too.

I was working for the leading print marketing company that was starting to have serious competitors and post-IPO we were really concerned that people were going to order business cards from us all the time (since they were almost free) to track our daily internal sales via order number. I can't say that we ever had hard evidence that this was happening but we thought we had suspicious activity once so we changed it.

An aside, changing order number to a hash value can be a huge pain in a manufacturing environment when people have (mis)used it to for a variety of purposes beyond order identification.

If you like math and strategy of course it's fun (and I really don't see how someone who doesn't like these things could work as a quant). You're getting paid to use skills you have to solve puzzles essentially.

I am working on something quant related---basically the part of the industry spawned by "Composing contrats: an adventure in financial engineering" (http://research.microsoft.com/en-us/um/people/simonpj/Papers...).

I enjoy the domain, but also that I can use functional programming languages.

Is ito calculus really the extent of the math you need to know to go into quant??

When my ex-boss was hired as a quant, he didn't know any stochastic calculus, but he was a full math professor (specialized in algebraic geometry).

So during his interview, they spent half an hour deriving the Ito integral from scratch and then proving some properties of it.

It's probably like any coding interview. You'll get asked a bunch of things that won't be a problem when you're on the job.

Basically, it's a "have you heard of this" type thing.

Oh well that makes sense, if stochastic calculus is all you need to know, i was half considering becoming a Quant there for a second ;)

For pricing, yes basically

Reply on news.ycombinator.com