# Master Thesis subject

**URL:** <https://forum.numer.ai/t/master-thesis-subject/6482>\
**Category:** Data Science\
**Created:** [June 21, 2023, 10:24am UTC](https://forum.numer.ai/t/master-thesis-subject/6482 "2023-06-21T10:24:03Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![thornam](https://avatars.discourse-cdn.com/v4/letter/t/eb9ed0/32.png) [@thornam](https://forum.numer.ai/u/thornam)\
**Post date:** [June 21, 2023, 10:24am UTC](https://forum.numer.ai/t/master-thesis-subject/6482/1 "2023-06-21T10:24:03Z")

</div>

Hi Everyone

I’m about to choose a Master Thesis subject and thought that you guys might have some great ideas for what to do a thesis about

To give some context, then I’m doing a Master’s in financial econometrics and therefore am more familiar with time-series modelling such as multivariate GARCH than with more straight forward Data Science. However, I have also taking courses within the Data Science field and am in general interested in DS (which is why I participate in this tournament), so I’m not affraid of taking on a subject of such. All ideas are welcome

I have already some subjects in mind, which are:

- Transformer based models for (multivariate) time-series forecasting as inspired by the referenced papers in this article ([Yes, Transformers are Effective for Time Series Forecasting (+ Autoformer)](https://huggingface.co/blog/autoformer)) and other recent papers within this field.  
(Much credit to @surajp for sharing this article on discord!)

- (Multivariate) Quadratic Hawkes processes for estimation of stock price distributions as suggested by Jean-Philippe Bouchaud ([[2206.10419] Multivariate Quadratic Hawkes Processes -- Part I: Theoretical Analysis](https://arxiv.org/abs/2206.10419))

- Or some less thought out subjects as Reinforcement Learning in a portfolio optimization framework, or Principal (Eigenvalue) portfolios as suggested by Bryan Kelly ([Principal Portfolios by Bryan T. Kelly, Semyon Malamud, Lasse Heje Pedersen :: SSRN](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3623983))

But nothing is locked down yet, so I’m very interested in hearing your thoughts. You might have some other interesting take on some of the subjects or sitting with an idea for a whole other subject that could be very interesting both for the Tournament/Signals or just in general

So I’ll appreciate all inputs ✌

---

<div class="post-metadata">

**Author:** ![surajp](https://yyz1.discourse-cdn.com/flex009/user_avatar/forum.numer.ai/surajp/32/2961_2.png) [@surajp](https://forum.numer.ai/u/surajp)\
**Post date:** [June 21, 2023, 1:28pm UTC](https://forum.numer.ai/t/master-thesis-subject/6482/2 "2023-06-21T13:28:54Z")

</div>

Nice. Mine was focused on self supervised learning + Custom loss functions to get robust embeddings.  
All the best!

---

<div class="post-metadata">

**Author:** ![thornam](https://avatars.discourse-cdn.com/v4/letter/t/eb9ed0/32.png) [@thornam](https://forum.numer.ai/u/thornam)\
**Post date:** [July 5, 2023, 8:29am UTC](https://forum.numer.ai/t/master-thesis-subject/6482/3 "2023-07-05T08:29:41Z")

</div>

That also sounds nice. Did you get any interesting findings?

Im currently working on some self supervised learning for the Numerai data just as a side project, and could be interested in reading yours if that’s possible