Quant Research

A good breakdown is here and the same author's PDF floating around from 2011 on algorithmic trading that he was forced to abandon upon signing NDAs to work where he is now. A good QR is an experimental scientist who has mastered the art of data analysis such as minimizing all sources of bias and can find a signal in a bunch of noise. This is all learned on the job. They are also salesmen (see below about explaining/presenting strategies). They are also responsible for optimizing the entire trading platform to reduce fees.

I have witnessed most of their time is spent explaining strategy for example they have some specific domain knowledge, use it to find a signal or optimization, and now have to explain and replicate it so others can adapt or sign off on it. The longer the strategy the more planning and meetings.

While they are spending time explaining there's a ton of other work to do that's where you come in as the lackey assistant or junior/trainee.

You can easily spot the shit work from the above blog post that a professional scientist making 7 figures would love to pawn off on some eager trainee. Even if there is a small army of teams set up for the QR department there is always something tedious they would rather pay someone else to do for them.

Shit work to them is basically anything that isn't directly involved in finding alpha. Maintenance of the research infrastructure is shit work and especially any kind of software writing and there's always a ton of software to write, visualizations to create or reports to file.

Why would you offer to do this? Because then they have time to show you the magic.

I did it. If you sell yourself and they want you around they will ask management to sign off and then you are presented with some additional non-disclosure agreements. Even in QR everything is compartmentalized it seems nobody ever has the keys to everything so if it's a big company they'll be willing.

The need for more researchers is never ending and where I work goes to enormous lengths to try and convince smart students to do this work by throwing massive sums of money at them.

Any linear/matrix algebra not covered in vector calc or data analysis is applied VMLS type 'least squares'. Really all you need is Taylor-series expansion, knowing what a derivative is, double integrals, and changing the order of integration to take that probability book.

I assume you can teach yourself all of the above and there's no need to go through it here instead I will do something completely different you may wish to take if you want to learn probability w/measure, are some kind of freelance quant who is interested in all those prediction markets, sports betting, foolish crypto speculation or other kinds of risky martingaling (gambling).

The free books here are 'DRIPPED' meaning the Riemann integral has been dropped and replaced by a simple teaching integral and these books contain very thorough presentations I've never seen elsewhere such as all the interpretations of a derivative, all the tests for continuity, all the definitions of a limit, etc. These texts were written by Professor Emeritus Andrew Bruckner and his wife Judith who are still active researchers despite being 90 years old and Brian S. Thomson who argues we should modernize the curriculum from the 19th century like dropping both the Riemann integral and unnecessary setups like Rolle's Theorem.

I will now attempt to learn measure theory using the free books Real Analysis [RA] by Bruckner and Thomson and Measure, Integration, & Real Analysis [MIRA] by Axler. We have YouTube and some functional analysis lectures from MIT to help us. If you don't care at all about game-theoretic foundations and want to learn probability as defined by measure theory this is your chance too. If you have only a highschool education, who cares, just try it first. This math is nothing like undergrad anyway.

Axler says we are supposed to be familiar with everything in this supplement. If you've never seen set notation look it up on YouTube.

Lightly review this material then refer to it later as it comes up. Any proof he shows is usually an example of the strategy to prove all the rest of that topic. For example use proof 0.2 to prove 0.3 we want an additive inverse so that a + (-a) = 0 so 1a + -1a = a(1 - 1) = a0 = 0. Since that is true, then (-1)a must be the additive inverse of a or -a.

Definition 0.23 for Dedekind cuts is well explained by Wildberger who thinks it is a bullshit abstraction but he gives a good presentation there what that definition means.

Here is a visual explanation of the Supremum and Infimum though the Axler definition makes it clear what it is.

TODO

---

Home