Forums  > University  > Master research paper  
     
Page 1 of 1
Display using:  

AK98


Total Posts: 3
Joined: Dec 2019
 
Posted: 2019-12-04 10:49
Hello,


I'm a master student pursuing a double degree program in Financial Economics / Applied mathematics and to get my diploma i should write a Research paper. 3 subjects has been proposed to me :

-Statistical arbitrage and pairs trading
-Factor selection in Finance
-Comparing discretization schemes in one-dimensional partial di􏰄erential equations for option pricing


Hence i wanted to ask which subject is more interesting and useful to my career knowing that i want to break into trading (not specifically HF trading) ?


Azx


Total Posts: 43
Joined: Sep 2009
 
Posted: 2019-12-04 11:23
All three were only relevant decades ago and have been done to death already. Why not try to do something more original if trading is really the career you want to pursue?

doomanx


Total Posts: 76
Joined: Jul 2018
 
Posted: 2019-12-04 11:31
@Azx while you might be right that these are common, doing a master project on the subject is a good way to learn the topic. There's also room to do something crazy even if it doesn't work/might not be a good idea, and the experience getting hands on with some data will be valuable.

The pairs and factor analysis topics link together quite nicely. Have a look at the Avellaneda Statistical Arbitrage paper to get an idea of how these two things link together. You have freedom to do something a bit wild for a Masters project, so how about you try to find factors using some cool new dimensionality reduction method. This could be a PCA variant (one of the many flavours of sparse/robust), a VAE, ICA (some trendy papers in nonlinear ICA going round at the moment).

did you use VWAP or triple-reinforced GAN execution?

AK98


Total Posts: 3
Joined: Dec 2019
 
Posted: 2019-12-04 11:58
@Azx Thank you for your reply, Actually because i'm in a special program i have to do 2 projects, Hence the professor suggested that i do a research paper on a subject of a normal difficulty so i could learn about it more as @doomanx said, then i'll be working on a serious "wild" project that will be the main master thesis.
Anyway thank you for your reply and please if you have any suggestion for a more serious relevant subject that i can do either for the project or the thesis i will be more than thankful if you share it with us ? Thank you very much.

AK98


Total Posts: 3
Joined: Dec 2019
 
Posted: 2019-12-04 12:01
@doomanx Thank you very much for the tips i'll be looking at the paper then i'll try to see if it suits the professor in charge.

doomanx


Total Posts: 76
Joined: Jul 2018
 
Posted: 2019-12-04 12:20
@AK98 I mean the topic is still extremely relevant (statistical arbitrage is one of the most common quant strategies) so to that end it's honestly not a bad project. The specific arb presented in that paper will be hard to pull off as it's probably priced in now, but the idea of finding a relationship that holds statistically and betting on deviations from the normal is something fundamental.

If you get interested in the dimensionality reduction topic you could take a look into the more 'cutting edge' methods like Diffusion Maps, maybe through the lens of multivariate time series.

did you use VWAP or triple-reinforced GAN execution?

Rashomon


Total Posts: 214
Joined: Mar 2011
 
Posted: 2019-12-15 01:59
None of the above. If you want to do HF look at the limit order book literature and start something from there. Also learn some networking and EE. Specifically you should be able to parse a PCAP file, understand UDP, and (easier) work with TAQ data.

On factors read:
- P Z Maymin’s thesis
- Eric Falkenstein’s book
- AQR’s recent results (plus their old papers)
- Douglas Bates’ now very old R-fin presentations on factor pricing


(But actually don't read them: just move on.)




Start with 3 blogs for HFT: mechanical markets, sniper in mahwah, and meanderful (twitter.com/sumohurd) for intro to HFT space.


Within last 10 years there have been plenty of papers on limit order book dynamics, and there was a quant-cup about writing a matching engine. These can be a better starting point. But, it might take longer to absorb these materials than you have to pick a topic. If you're desperate on time, there are a couple microstructure textbooks which you can quickly copy and iterate on a paper. Better is to read CME/BATS circulars, or KCG order-type papers, and find a tiny, tiny question in there. If you had another half year or year to prepare the question, one could probably set up a research project that's likely to work and answer a question which can be stated in plain English and doesn't only relate to previous models published about.

Maybe the easiest is go to MIDAS or WRDS and grab some TAQ data. Anything you do will practise your basic facility with the computer, which is necessary to be able to actually do anything helpful for an employer. Firms don’t need more mathematic masters; unfortunately you should have read more before enrolling. As far as “breaking in”: many other mathematic students had that same idea, but don’t have the basic computer skills to do useful work.


Also, why not pick something besides US large-cap equities? It’s a cheap way to stand out, a little.




HTH.

jamescameron


Total Posts: 4
Joined: Apr 2019
 
Posted: 2020-01-07 21:03
Writing research paper plays an important role in higher education life. Most graduates and postdocs spend much time on writing research paper. Fdcpa attorney
Previous Thread :: Next Thread 
Page 1 of 1