Finance
Vice President at BlackRock, London, Jan 2020 – Jun 2020 Associate at BlackRock, London, Sep 2016 – Dec 2019 Analyst at Standard Life Investments, Edinburgh, May 2015 – Mar 2016
Upon the end of my PhD study, I received an opportunity to work in Standard Life Investments as a quantitative researcher. After eight years training as a mathematician at university, I was very keen to apply mathematical techniques to solve real-life problems in industry. With great curiosity and enthusiasm, I started my journey as a quantitative researcher in finance.
My main responsibilities as a quantitative researcher included : – portfolio construction, trade and risk analysis for multi-asset portfolios – the research and development of risk factor models for equity portfolios. I gained experience in building mathematical and statistical models for pricing financial derivatives, predicting price movements of financial instruments, exploiting investment opportunities and capturing the uncertainty of the financial markets.
In various projects that I have conducted, I used statistical and natural language processing (NLP) techniques to extract useful information from a variety of structured and unstructured data. I analysed the relevance and significance of environmental, social and governance (ESG) metrics with respect to the risk and stock return characteristics of investments. I studied the economic impact of the interconnectedness between companies in a variety of casual networks including supply chains, lending relationships and geographic location. I proposed and implemented a hierarchical mixture of experts (HME) machine learning model for forecasting equity idiosyncratic risk.
As a quant researcher, I worked with numbers and code in my day-to-day job. However, one of the most important things I learned is that investment management is a social science, and questions in this field can not be answered in a numerical way without understanding of other social activities. Successful applications of mathematical models must be developed with profound understanding and insightful intuition of finance and economics.
“I can calculate the motions of heavenly bodies, but not the madness of people.” — Sir Isaac Newton
Making good investment decisions requires good understanding of stock price movements as reflection of trading activities which are closely related to politics, economics, geographic, culture, company operations etc. With an objective to forecast return, fundamental analysts study company fundamentals, economists study politics and economics, and portfolio managers combine different views from these studies when constructing portfolios. Guided by the intuition that risk and returns of stocks are driven by some risk factors, fundamental factor risk models (see Connor (2010), Bender and Nielsen (2012)) have been developed and commonly used for forecasting risk and constructing portfolios.
A fundamental factor risk model decomposes equity returns into country, industry and company characteristic and idiosyncratic factors. To refine this building block for risk and return decomposition, a substantial research effort has been spent on constructing factors and finding mapping from factors to stock returns. A variety of mathematical tools have been applied and developed for these purposes: multivariate regression, lasso regression, stepwise regression, Vasicek beta adjustment, instrumental PCA etc.
“Do what is right, not what is easy nor what is popular.” — Roy T. Bennett
Although traditional economic theory (see Friedman (1970)) suggests that there is no environmental or social responsibility for business to make profit, incorporating ESG aspects into the evaluation of business has recently attracted considerable attention in the industry. While ESG investing mandates require excluding corporations associated with undesirable ESG practice and policies, challenges emerge due to voluntary discloser biases in many ESG metrics and a lack of consensus on precisely what is being measured.
Meanwhile, the advancement of mathematical techniques, big data processing tools and NLP models has given rise to building models to extract information from text data. Hoberg and Phillips (2016) suggested using corporate business description to construct a product network, which places companies in a space based on the products they develop, so that companies developing similar types of products are gathered together. In addition to giving timely updates for company exposures to sectors, this tool has many interesting potential applications.
To address challenges in ESG investing, such as the voluntary discloser biases in many ESG metrics, one can develop an E, S, or G network using data harnessed from less obviously biased sources of data, such as company filings, news articles, conference call logs, corporate websites, supply chain relationship, management disclosures etc. Such models, developed with discretion and expert judgment, improve the allocation of social research efforts and judgment.
Conducting research was but one element of my job. I enjoyed writing papers and articles, conducting university lecturing and client presentations, and participating in charity events. I valued the opportunities working collaboratively with many different teams, from initiating a research idea, to overshooting hurdles in research, to productionising the research outcome. I worked with supportive managers who showed the meaning of leadership, who helped me to grow my technical skills and more importantly, to grow as a better person.