The financial markets industry is currently looking for fresh talent to develop new products, services, and financial models in line with the economic and technological upheavals: high-frequency trading, blockchains, cryptocurrencies, decentralized finance, and many others.
ESILV students get hands-on experience working on financial engineering projects aligning with the latest technology trends and their applications in financial markets. Spotlight on three engineering projects completed by graduate students as part of their Master in Engineering curriculum.
Quant trading: pattern strategies design, machine learning approach
The objective of this PI2 project was to develop an algorithmic trading and execution strategy for Systemathics. As a team of 4 ESILV students, we aimed to implement a peer trading strategy using statistical data and Machine Learning models and to analyze our results in order to deduce future improvements. By definition, peer trading (or pair investing) is a trading method that is indifferent to market conditions and fluctuations.
In order to implement a complete peer trading strategy we have defined and completed several tasks:
– The first is the selection of assets. For this several models are possible, the goal being to choose stationary and/or correlated asset pairs
– The second step is the construction of the strategy for the chosen pairs (when to buy, sell, what amount…)
S&P500 co-index replication based on co-integration
The topic deals with US index replication via a trading strategy based on cointegration, trying to improve it if possible so that it is a successful long-short strategy. We refer to the article “The Cointegration Alpha: Enhanced Index Tracking and Long-Short Equity Market Neutral Strategy” by Carol Alexander and Anca Dimitriu published in 2002.
This article presents 2 applications of this strategy: a classic index-tracking strategy and a long-short equity market-neutral strategy. Here the portfolio optimization is based on co-integration rather than correlation, unlike other traditional strategies. The first strategy aims to replicate a benchmark index accurately in terms of return and volatility, while the other seeks to minimize volatility and generate consistent returns under all market circumstances. Thus, we look for several combinations of these two strategies to find the best combination.
Building a Python cryptocurrency robot
Cryptocurrencies are becoming more and more popular, and their use is becoming more democratic. They are now more and more numerous, and easy to use. However, earning money via these crypto-currencies is not so obvious. Many strategies already exist.
The goal of our project was to write trading bots in python that perform on the crypto-currency markets. Specifically, the objectives were: to define key indicators, define trading strategies, implement them in python, backtest our strategies in order to improve them, and at the end of our project, deploy our code live.
In this project, we tested different strategies “The momentum trading strategy”, “The mean reversion on pairs”, and “The Parabolic SAR”. We compared them and tried to determine the most effective. We ended up deploying the mean reversion strategy live.