Monash Internship Program
Optimal Decision-Making for Major Portfolio Investment (August 2023)
Interns: Weiyao Cheng, Denwang Li, Wei Yun Shen,
Supervisors: Dr Blessing Taruvinga, Dr Geoff Lee, Dr Ying Guo, Dr Jiaming Li and Dr Zili Zhu
Scope: The objective of this project is to establish a dynamic decision-making process for investing in two major assets, and only one asset can be chosen at any of the future multiple discrete time intervals, the decision for not investing or delayed investment is part of the decision process. For this project, numerical techniques often adopted in the valuation of American options or Bermuda options can be utilised (such as the least-square Monte-Carlo method), and we envisage machine-learning techniques can also be explored to enhance computational efficiency.
The output from this project can be applied to major investment decision-making often encountered in mining, agriculture, and major infrastructure, in addition to finance.
Forecasting economic variables in the US market (February, 2023)
Interns: Jiacheng Gao and Yihan Chen
Supervisors: Dr Blessing Taruvinga, Dr Geoff Lee, Dr Jiaming Li and Dr Ying Guo
Scope: The project will focus on using economic data within the US environment for: 1) building a multi-variate regression model to forecast an economic variable from input multiple economic variables, 2) deploying Machine-learning algorithms to identify clusters in these economic variables. For 1), the first test example can be to forecast CPI and to identify suitable major economic drivers as input variables. A second example for 1) can be to forecast a commodity price (e.g., copper price) and to find the corresponding main drivers as input parameters of the regression model. For 2), the project will need to experiment with different classification techniques for generating clusters with more explainable features.
Identifying relationships among forecasted economic variables (August 2022)
Interns: Lijun Ma and Haitong Zhu
Supervisors: Dr Ying Guo, Dr Jiaming Li, Dr Geoff Lee
Scope: From large number of forecasted economic variables, it is possible to quantify and identify correlations among these economic variables. Importantly and conversely, these quantified relationships can also be used to assess the validity of the forecasting model. This project will first focus on quantifying the relationship among the economic variables through Machine-learning techniques such as clustering methods. Secondly, if the project proceeds well, we can use the identified relationship among the future values of forecasted economic variables to modify the structure of the forecasting model so that the identified relationships are aligned with established consensus.
The forecasting model will initially be 1) simple regression model, 2) the cascading SUPA model. The Machine learning techniques for time series clustering will primarily be the k-means clustering technique.
Machine-learning for financial option pricing under stochastic volatility models (August 2020)
Students: Ruiyuan DONG, Zhou Xu
Supervisors: Drs Geoff Lee, Jiaming Li, Blessing Munhumwe,
We implement Machine-learning (ML) techniques/algorithms to price financial options. The objective is to investigate if ML can price options faster than conventional option pricing methods such as Monte-Carlo and/or finite-difference methods. The project may utilise different ML techniques, such as modules available in Python (such as PyTorch). For example, the deep neural networks (DNN) can be trained to reproduce precomputed option prices, then, the trained ML algorithm can be used to generate option prices for any given new model parameters. The pricing error of using ML algorithms should be the focus of study. The Black-Scholes model should be used as a first test case. If successful, a more sophisticated stochastic volatility models such as the Heston model will be the second test case.
Risk Measures in ASX200 Trading Data via Machine Learning and Statistic Techniques (August 2020)
Students: Danyang LIU, Lan SHU, Jie XU
Supervisors: Drs Dennis Sams, Ying Guo
Using existing trading data of individual stocks and the reconstructed ASX200 index, this project will focus on implementing PCA, Machine-learning techniques (e.g. clustering) and factor modelling to identify risk factors and establish risk measures for the time series trading data of ASX200 in terms of sectors, serial correlations and market caps. The analysis can extend to cover the top 20 or 50 companies of the ASX200. This project will leverage the outcomes of two earlier projects in which complete ASX200 index data were generated for the time periods of COVID-19 and GFC, and Machine-learning algorithms were implemented to identify risk features. The objective of the current project is to develop risk measures for the ASX 200 trading data from 2001 to 2020.
Superannuation in China (August 2020)
Students: Jingran WANG, Qiuyan XIE
Supervisor: Drs Zili Zhu, Monica Chen, William Szuch,
This project will focus on identifying and establishing optimal accumulation strategies to maximise pension entitlement and total wealth upon retirement for a typical working person in China. Another emphasis is on the study of the various wealth management products including life insurance products that are commonly adopted apart from pension savings. The first objective is to convert the risk profile of a typical working person to create a corresponding portfolio with the appropriate risk measure such as volatility, and/or drawdown limit. Once the mapping of the personal risk profile and portfolio risk is established, existing simulation models already developed and constructed can be used to forecast the future outcomes for the expected pension entitlement and potential accumulated wealth. Finally, the analysis of pension system in China will be compared with the superannuation environment in Australia.
Australian Equity Performance During the COVID-19 Pandemic (ASX 200 size and sector index construction and analysis) (Feb 2020)
Students: Mr Runar Oesthaug and Ms Jiaqi Pan
CSIRO Supervisors: Dr Dennis Sams, Dr Zili Zhu, Dr Geoff Lee
Using Bloomberg trading data to build S&P ASX200 into a SQL database, we focus on implementing various analysis on the market performances during latest COVID-19.
Kernel differentiation for exotic option pricing and hedging: (Feb 2020)
Students: Mr Brishav Hazarika and Mr Sarab Monga
CSIRO Supervisors: Dr Nicolas Langrene, Dr Wenfeng Dong
This project focuses on stabilising the Monte Carlo pricing of exotic options by the so-called kernel differentiation technique. Kernel density estimation can be used as a variance-reduction technique for estimating option prices and Greeks. The aim of the project is to implement and test this statistical technique on a series of options in order to identify the best combination of kernel shape and bandwidth parameter in the context of financial option pricing.
Study of SSE Composite Index attributing market return to individual stocks/and or groups of stocks. (Feb 2020)
Students: Meihan Liu / Yuxin Deng,
CSIRO Supervisors: Dr Dennis Sams, Dr Geoff Le, Dr Zili Zhu
The focus is on the structure of the SSE Composite index and its market composition including links to specific sector and company data. Part of this project is to identify the drivers of returns of the SSE Composite Index and to analyse how it has performed (through proxy sectors and/or companies) during the recent crisis (GFC and the current coronavirus crisis). An interesting analysis will be to contrast the SSE Composite Index performance with Australian ASX200. The ideal outcome of this project is to link the analysis of this project with an existing RiskLab intern project which is well advanced on analysing the ASX200 to generate the contrasting features during the market crashes for the SSE and ASX200.
Analysis of Future Superannuation Outcomes in China and Australia (Feb 2020)
Students: Yingnan Wu & Congying Wang
CSIRO Supervisor Dr Monica Chen, Mr Bill Szuch, Dr Zili Zhu
This project will first analyse the current status of the superannuation system in China with regards to the government pension and typical private superannuation saving practices. The project will then focus on utilising existing economic data to forecast future economic scenarios by calibrating a VAR(1) model or the CSIRO SUPA model (if the required Chinese economic data is available). These forecasted economic scenarios will be the basis to project future retirement saving outcomes upon retirement of a typical working person in China. The project will present and compare the retirement systems of China and Australia.
Risk Factors and Clustering in ASX200 Trading Data via Machine Learning Techniques (Feb 2020)
Student: Xin Sun & Chengda He
CSIRO Supervisors: Dr Wenfeng Dong, Dr Nicolas Langrene
Using existing trading data in the SQL database we have built up, the project will focus on implementing simple Machine-learning techniques to identify risk factors and/or clusters in the time series of ASX200 data series and the top 20 companies of the ASX200. Techniques from pandas/sklearn packages can be applied for stock index clustering for comparison with conventional sectors defined in the ASX200. Exploration of trading pair strategies can be performed if the first stage of the project progress smoothly.