Projects

A Parsimonious Stochastic Simulation for LOB Dynamics

We are committed to draw predictions of the midprice of BTC based on half-day data. There are many statistical methods for features engineering and draw predictions based on the extracted factors, while, it performs poorly for longer period prediction. In this project, we put forward both basic statistical methods and stochastic methods, and make comparison between them. Our stochastic approach based on Cox Process is much more robust for longer period prediction and is simple to implement.

Analysis towards the EU ETS markets and related commodities, Part A - Mechanism, Statistical Insights and Pricing

This report would cover the mechanism of Emission Allowance, based onwhich we would propose correlation analysis of EU ETS with other derivatives or commodities in energy markets. We further propose the pricing theory of EU ETS to predict the price dynamics. The seasonality analysis is undergone as part of price dynamics analysis.

Analysis towards the EU ETS markets and related commodities, Part B - Time Lag, Correlation and Trading Strategy

This work goes deeper in the correlation between fuels and the EU ETS price. We discovered time lag behavior, specific correlation between different commodities with the EU ETS price and based on this, we could come up a strategy to trade. We did many adjustments in order to get hold of such correlation.

Analysis towards the EU ETS markets and related commodities, Coding and Simulation Part: Pricing and Prediction based on equilibrium Fuel Switching Price

This work is inspired by the work of Rene Carmona et al, SIAM J. Control Optim., Vol. 48, No. 4, pp. 2168-2190. We tried to simplify the framework and fit it to the real trading scenario. The codes are efficient and results match the mechanism in reality.

Introductory work on Spectral Theory

Operators have their spectrum. The spectrum reveals critical features of the dynamics, such as frequency and energy. This project combines theories with practical implementation with regards to the spectral theory. This could be extended to the stochastic world, to describe the characteristics of random systems, such as stocks, with high efficiency.

Signature Method in Analysis and Prediction of Stock Price

We consider the potentials of signature method for predicting and analysing the stock price. We considered its computational efficiency and usage under complicated scenarios. Moreover, this work casts more in-depth ideas that reveals fundamentals of signature method.

Analysis of Tick Data for High Frequency Trading

This is an interesting test from Itarle. Here is my attempt, where we are concerned about the statistics, and round effect. In this trail, I compared the difference of mean and median, and also created measurement for different purposes. The work is based on C++

Kuan Yang