Most important papers for quantitative traders | Hacker News Reader

Most important papers for quantitative traders

qmr.ai

185 pointsAnon8461 comments

Quant trader here... I'm a big seller of this list. Making money tends to be a relatively empirical endeavor. It's all about having information about the future and using that in judiciously way and less so about any particular theory or model. I see someone else mentioning Grinold and Khan "Active Portfolio Management", I can't recommend it enough, it's basically a how to for making money quantitatively in a principled way, there are lots of "tips and tricks" that go on top of this and it really helps to have some good intuition for the space you are trying to operate in (by that I mean understanding the eigenvalues and eigenvectors of your risk matrix). T-costs are also extremely important and the main "enemy" it's trivial to make money if you don't have to pay to trade.

Steven Boyd at Stanford and his students / colleagues are probably the richest seam of up to date portfolio optimization wisdom. If you are using python you shoult probably be using CVXPY to build your portfolio. He has lots of good papers, e.g. see [2].

Of course you also need an "edge", that information about the future, and that's the jealously guarded part...

[1] https://books.google.co.uk/books/about/Active_Portfolio_Mana...

[2] https://stanford.edu/~boyd/papers/pdf/cvx_portfolio.pdf

Seller of this list, as in you think it's not a good list?

Not really going to help you make money...

A big seller? So you are shorting this list?

Little know fact: the CAPM is an equality of random variables, not just their expected value. https://arxiv.org/abs/2009.10852

> having information about the future

How, uh, do you do this legally?

>If you are using python

And if you're using C++?

> How, uh, do you do this legally?

It's simple to have information about the future in entirely legal ways. Usually the future information is available, just unequally distributed.

The best example of this is the movie "The Big Short" where it the information about the upcoming crash of subprime-backed bonds just required people to bother reading large amounts of bond composition documents. Only 3 groups really did this.

Another good examples is how some funds pay for logistics intelligence (via satellite reconnaissance, customs declarations etc) to forecast sales figures.

> How, uh, do you do this legally?

I don't think you can generally, but in specific contexts, having a model of the system allows you to extrapolate. For example, people who foresaw how the pandemic would or even could play out made lots of money. The further out you can connect the dots, to secondary or tertiary effects, the better. I did nothing personally.

Forecast! Time to reach for your linear regression, GBDT or NN.

> How, uh, do you do this legally?

Why would it be illegal if you're not directly involved with the corporation? Surely insider trading implies actually being some kind of insider. Like some politician selling stock before some regulation takes effect.

Trading on insider knowledge is illegal afaik

Seems vague to me.

So you go to an event or something. Company guy says something stupid that convinces you they're doomed to fail and then you make money by shorting their company's stock. That's illegal?

Well, maybe. I think it depends.

Certainly you can't have your friend Bob who works at Acme tell you stuff then suddenly you're in the clear. That isn't how it works.

But if let's say Bob does something dumb that leaks info publicly, sure - it's public now. You can trade on it.

I'm far from a lawyer but there's nuance here.

You know rates will go down eventually. You know the debt ceiling deal will probably pass and the market will like it. Plenty of things we can know.

Hey there! I'm the author of the article. I just arrived here because I saw a crazy uptick on google analytics. I'm glad most of you liked the article.

Keep up the great content :)

I found other impactful and more recent papers via MirrorThink.ai that discuss various aspects of quantitative finance, trading, optimal execution, energy prices, GARCH, option valuation, portfolio selection, Kelly Criterion, Capital Asset Pricing Model, optimal trading signals, Efficient Market Hypothesis, Black-Scholes model, and market overreaction. Here are some key findings from these papers:

1. Portfolio Optimization-Based Stock Prediction Using Long-Short Term Memory Network in Quantitative Trading (Published on 2020-01-07) - This paper discusses the use of Long-Short Term Memory (LSTM) networks in quantitative trading to minimize risk and maximize return based on historical performance. It highlights the benefits of quantitative trading, such as lower commissions, anonymity, control, discipline, transparency, access, competition, and reduced transaction costs.

2. A Markov-Switching VSTOXX Trading Algorithm for Enhancing EUR Stock Portfolio Performance (Published on 2021-05-02) - This paper presents a Markov-switching trading algorithm that uses the VSTOXX index to enhance the performance of a EUR stock portfolio. The algorithm is based on the mean-variance portfolio selection, which aims to maximize the Sharpe ratio.

3. Price discovery in the cryptocurrency option market: A univariate GARCH approach (Published on 2020-08-31) - This paper applies two different GARCH processes to Bitcoin and CRIX, showing that the GARCH(1,1) option pricing model provides realistic price discovery within the bid-ask prices suggested by the market.

4. The Capital Asset Pricing Model (Published on 2021-09-03) - This paper discusses the evolution of the Capital Asset Pricing Model (CAPM) and its connection to behavioral accounts of evolutionary asset pricing, segmented markets, multifractality, and the fractal market hypothesis. It highlights the importance of considering heterogeneity among investors and the implications for the efficient market hypothesis.

[1] https://doi.org/10.3390/app10020437

[2] https://doi.org/10.3390/math9091030

[3] https://doi.org/10.1080/23322039.2020.1803524

[4] https://doi.org/10.3390/encyclopedia1030070

Probably better to learn the basics and get a good overview, eg by reading Grinold & Kahn instead:

https://books.google.com/books/about/Active_Portfolio_Manage...

Went looking for this on your recommendation and found that there is an updated version/follow-on text, linking here for anyone else that may find it useful:

https://www.google.com/books/edition/Advances_in_Active_Port...

I stopped at the table of content but the bibliography covers what any introductory finance 101 course would cover. I interpreted the title as suggesting there was a bit of novelty in there so it's a bit disappointing.

Without the above papers you cannot invest while claiming doing anything else than playing at a casino. But it's clearly not sufficient to design a profitable quantitative strategy in 2023.

>Without the above papers you cannot invest while claiming doing anything else than playing at a casino.

I don't think that's a fair statement, although I agree with the overall sentiment. Maybe the right term instead of "invest" would be "actively trade". Putting a chunk of change into long-term positions (especially stock) on large profitable companies as well as indexes and dividend-generating equities with a view towards cashing out in 30-40 years (and semi-actively monitoring said portfolio) isn't really the same as playing at a casino. If I'm looking for a 10-30% return in a day or a week, yeah, that's playing at a casino. If I'm looking for 7-10% a year, that's just me protecting my money against inflation.

Yeah these are classic quant finance papers, but all in all not super helpful for anything except a basic foundation.

Thanks for sharing! My list of 12 most useful resources for machine learning quants can be found here: http://gokhanmergen.com/quantBibliography.html

This list was compiled in 2009 before I took a full time job in an algorithmic trading company, but it's still relevant :) If anything ML is more relevant than ever in trading, except perhaps Deep Neural Nets, Transformers, Large Language Models etc are the norm today.

Kinda tangential, but hopefully someone lurking in here will bite:

With the growing popularity of passive strategy among institutional and retail investors, will EMH break down and create opportunities for active strategy again? As I understand it, active strategy is a borderline fools' errand on the timeline of ten or more years. But if everyone just buys the S&P, surely that means fewer eyeballs on price discovery and more pricing inefficiencies, no?

There has always been room for active strategies, and there will continue to be room for active strategies. It's very likely _your_ active strategy isn't as good as buying and holding an index fund.

Table of contents? Blast from the SEO spam of the past.

Snark aside, very decent bibliography for the intended audience: independent traders who are building automated trading programs for their personal accounts.

Nice article, I'll definitely read some of the outlined books. Thanks for sharing.

My personal experience is that you don't need to fully understand the Black Scholes Pricing model in order to trade profitable options.

As an example consider the public income trades, such as NetZero, Boxcar, M3, Theta Engine. Trading those doesn't require you to understand how Implied Volatility.

One can argue, however, that selling options and hedging them isn't the 'Quant way' of profiting from options.

The Black Scholes paper is actually pretty difficult, and not terribly rewarding with respect to developing intuition. To develop the intuition, get a full handle on put-call parity, the construction of replicating portfolios, and then risk-neutral pricing. Additionally, always have an eye on the intrinsic value, time value and insurance value of options and how they move with respect to the options characteristics.

I found learning about the binomial model to be particularly good at establishing intuition about option pricing. Nowadays it's considered a toy model but it makes the ideas of a replicating portfolio and risk neutral probabilities very clear with nothing more than basic high school math.

I am not a quant but I am working for one. Am I wrong to think that the quant analysts or "scientists" are not actually figuring out something fundamental about the market? You win by being "more complex" than your competition, which in turn makes the whole market more complex and this goes on endlessly.

Yeah I think you are wrong, and quants don't actually think like that.

> quants don't actually think like that.

Help me understand it then. AFAIK, this is how it goes:

1. The market can be modeled using A/

2. Someone figures out it can be modeled by A'.

3. The market inevitably changes because of this application of A'.

4. The real model for the market shifted to B.

5. Repeat 1.

Which just means that the model for the market will only get more and more complicated. And the way to win is to have more and more sophisticated model to capture every other model.

I recommend anything by Marcos López de Prado:

https://quantresearch.org/

He has written a book and published several articles on financial machine learning, including what to look for,how to avoid overfitting, in detail, and done so far better than I've seen elsewhere.

Good luck!

The guy might be smart and write popular textbooks, but he has not successfully managed real money over an extended period of time as far as I can tell. He "left" Guggenheim after his algos were basically found to be crossing trades with the firm, and he "left" AQR after it was determined that he essentially brought nothing new to the table.

This is what I wanted to do when I was getting my math degree! I wanted to be a quant. Things went a different direction and I'm a programmer now. Is there any hope for me? Think I could still chase it down in my spare time, or is it something I need, say, a master's degree for?

You will be better off going independent, than trying to break into the industry (unless your PhD was from a top program, and you're on the younger side).

In that vein, Ernest P. Chan's books will give you the toolkit to begin, and then you can figure out the rest as-needed. Quantitative Trading (2nd), Chan I believe is the first in the series. Algorithmic Trading I believe is the second. And Machine Trading is the last.

That's awesome, I appreciate it. Do I need a ton of money to start off? Is it better _not_ to have a lot of money to start off to create less trouble for myself?

I'd heavily advise not to do what they're recommending, and try out applying to trading firms if you want to break into the industry.

I work at a trading firm and a decent number of our quants were all just programmers beforehand with no trading experience. Additionally, there's no downside to applying to a bunch of them and seeing if you get an offer, but there's a HUGE downside to gambling your own money in a field you don't understand.

You could also do what I did and apply for a normal programmer role at a trading firm. Then you'll get some exposure to the quant world and if you want, you can try to pivot internally (I haven't done the last part and am not super interested, but plenty of people I know have).

> Do I need a ton of money to start off?

well the best way to make a small fortune is to start with a large fortune ;)

>Independent

Really? Everything I've ever seen recommends against this.

Don't do it if you need the money.

At that point, it's like any other hobby: horse-racing, sports-betting, or purchasing YCH commissions to hang up on your dining room wall.

Hi there! I'm the author of the article.If you have a programming skills and a mathematics background, it is possible to break into the industry. Even better if your current role is as a data scientist of machine learning engineer.

I work as a quant on a second tier hedge fund. The pool of potential firms is actually pretty big, but most people think that the only shops out there are citadel, two sigma or rentech. That is definitely not the case, and the salaries are still excellent (comparable to FAANG).

What are the noncompetes like? I've heard that once you work for one shop, you basically can't work/apply your knowledge & skills anywhere else, which leads to stress about keeping your job.

That's not true, NCs / gardening leave last about a year.

The fail rate for traders in a real trading firm is something like 50% anecdotally (i.e. 1/2 of those hired don't make past 12mo mark)

The fail rate for retail traders without the professional environment backing them would be >95%.

That is to say the best thing you could do to increase the probability of your success is to get in the door at a reputable place.

https://robotwealth.com/ is probably the only source of information for a retail trader that I'd recommend. It's still far inferior to actually getting a seat at a real shop.

There's plenty of work for programmers in trading companies too - much better job stability than for traders.

Stephen Ross (https://en.wikipedia.org/wiki/Stephen_Ross_(economist) )'s books on APT / CAPM are pretty good too imho.

Can you share the books that you are referring to? A quick search on https://www.goodreads.com/author/show/169424.Stephen_A_Ross didn't seem to turn up relevant results

Papers are pretty good https://en.m.wikipedia.org/wiki/Arbitrage_pricing_theory

The CAPM model and APT imho is what portfolio management theory is based on: valuing equities or other instruments relative to each other. Useful for pairs trading, alpha, beta, and really all risk management. His portfolio theory textbooks are good too I think.

Most serious trading strategies can be summarized by highlighting a sentence in Hull’s book. That’s how it was in the 00s anyway. It’s all just innovation was execution and not just speed. It relationships and little bits of edge on top of existing old strategies.

Stephen Ross actually seemed to provide the first real theory for risk management though. I wouldn’t be surprised if most prop trading firms are still based on that.

I cannot open the link

No problem here.

https://archive.is/DCm8e

Commenting to remind myself later

FYI, you can click on the "favorite" button to add it to your list

The author seems to imply that there's a Nobel Prize in economics. There is no such prize. There's only: Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel

What you mention is the Nobel Prize in economics, it's the same thing. It wasn't one of the original prizes set out in Nobel's will, but it is awarded by the same institution and follows the same procedure.

Only an absolute pedant who is looking to argue trivialities would bicker over the name of "Nobel Prize in Economics" vs. "Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel".

[1] https://www.nobelprize.org/prizes/lists/all-prizes-in-econom...

> absolute pedant who is looking to argue trivialities would bicker

Well, yes it's a more than a little pedantic, but I think your statement may be too strong.

For example, at least one member of the Nobel family has objected to using their name on the prize:

> Nobel accuses the awarding institution of misusing his family's name, and states that no member of the Nobel family has ever had the intention of establishing a prize in economics. [from Wikipedia]

Also, while it's true that they're administered similarly, it also is true that they are funded from different sources. The Economics prize is funded by the (100% state-owned) central bank of a monarchy. Does that matter? Maybe not, but it's certainly a bit smelly and some of their picks in the past don't seem entirely justifiable solely on academic merit.

> Only an absolute pedant who is looking to argue trivialities would bicker over the name of "Nobel Prize in Economics" vs. "Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel".

well this is HN after all

There is some value to being pedantic. I did not know this fact about the economics Nobel and now I do.

If you would establish a price and someone would pick up your name for another price, would that make you happy?

Alfred Nobel did not establish a price in economics. So IMHO this price should not have his name. Sure, the official name is a bit bulky. We can certainly find something more appropriate.

Tesla did not establish electric veichile company and Max Planck did not, in fact, erect a building in my university's square.

It's not a random person who picked up the name. Nobel established the various prizes in his will to be administered by certain committees and it is one of those committees that added the prize in economics to be awarded under the same criteria and at the same time as the other prizes.

It is definitely notable that that the prize in economics is not one of the original prizes and I would never argue that it is. But acting like there is no strong relationship between the two that one would claim that a Nobel Prize in economics just doesn't exist is I think misguided. There is such a prize, it is not one of the original prizes and does have a different history, but it is strongly related to the other prizes.

"... it is commonly referred to as the Nobel Prize in Economics." [0]

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

akshually....

Reply on news.ycombinator.com