A1: Thesis request

Informal thesis request sent by email:

Having a strong software engineering background, my aim is to build a product which involves research within the field of data science as well as some financial aspects -- since I find this combination rather exciting. My personal projects involved to build a rudimentary auto-trader that spots arbitrage opportunities among several markets. I further built a visualization of this technique during the data visualization course, see final product here: https://www.youtube.com/watch?v=kcM1eOcQkfg

Next, over the course of the neural net course at TU Delft (by Mrs. Redi), we trained LSTM neural nets to predict future Bitcoin prices (regression based). You can find the report attached. I have continued with this project, in fact I spent a good part of the summer break on this, as my goal is to build a machine learning based auto-trader. This comes with a long chain of engineering tasks which include for example to build a reliable data source, micro-service oriented software architecture, evaluating ML models and last but not least the trading strategy including the handling of trade execution and their reporting. Many of these tasks are meanwhile already implemented as part a framework.

My ambition is now to dig further into the research of the crypto-currency-price prediction, -strategies as well as -trade execution using machine learning techniques. Apart from the neural network research (LSTM time series), my interests include:

• Market making strategies
• Optimal trade execution
• Detecting latency arbitrage opportunities

A potential master thesis could therefore be: to recognize patterns in orderbooks to predict short term price movements using limit orders (see http://www.tandfonline.com/doi/abs/10.1080/14697688.2015.1032546) and thus execute optimal trades (see http://dl.acm.org/citation.cfm?id=1143929) in the most suitable market at the given time, according to latency arbitrage forecasts (no comprehensive research found yet).