CV

Download my CV (pdf)

Education

• PhD in Finance, The University of Manchester, 2025 (expected)
• MSc in Quantitative Finance, The University of Manchester, 2019
• BA in Economics (Finance), Shandong University, 2018

Research Interests

Empirical Asset Pricing, ESG (Corporate Social Responsibility), Machine Learning, Natural Language Processing, Behavioural Finance.

Job Market Paper

Intensity Bursts in WallStreetBets Discussion and Stock Markets Trading (With Yoichi Otsubo and Ser-Huang Poon)

• Abstract: Our study introduces the concept of WallStreetBets (WSB) submission intensity bursts (IB), offering a novel perspective by distinguishing varying levels of social media activity. Using a Difference-in-Differences approach, we establish a causal link between IBs and key market outcomes, including trading turnover, daily and overnight abnormal returns. We further find that IBs during regular trading hours (RTH) have a stronger and more lasting impact on stock returns compared to those outside RTH, with swift reversals observed on a half-day basis. Additionally, stocks targeted by WSB users typically exhibit a decrease in short interest ratios following an IB, suggesting the WSB users may anticipate short squeeze opportunities, adding strategic nuance to social media’s market influence.
• Presentations (*by co-author): Financial Econometrics Conference to mark Stephen Taylor’s Retirement, Sofia University, 2023 Nippon Finance Association Annual Conference, 12th International Conference of the Financial Engineering and Banking Society (FEBS), 2024 British Accounting and Finance Association (BAFA) Annual Conference

Publication

Hayakawa, K., Otsubo, Y., Poon, S. H., & Wei, S. (2024), ‘Vocabulary Herfindahl Index (Vocahin): Linguistic dominance and collective effervescence in WallStreetBets, Economics Letters 244, 112027.
(With Ko Hayakawa, Yoichi Otsubo and Ser-Huang Poon)

• Abstract: Our analysis of over 150,000 WallStreetBets comments reveals dominant phrases when members of the community experience collective excitement and solidarity (Collective Effervescence). Using the novel Vocabulary Herfindahl Index (VocaHIn), we find that stock returns volatility increases after CE, and vice versa.

Work in Progress

LLM and Transparency in Supply Chains (with Ser-Huang Poon)

• In this paper, we apply a state-of-art Large Language Model to evaluate Social reports related to business supply chains. We also explore a number of factors that are crucial for evaluating corporate disclosure quality.
• Presentations: 2022 British Academy of Management (BAM)

User Patterns in Financial Social Media (with Yoichi Otsubo and Ser-Huang Poon)

• In this paper, we implement Prompt Engineering, the process of structuring an instruction that can be interpreted and understood by a generative AI model, in specific financial social media text data. We also find an intriguing behavioural pattern that shows how the users on financial social media interact with each other.

Teaching Experience

Teaching assistant

• BMAN30091 Financial Derivatives (UG, 2021-2024)
  • Received 2022 AMBS Teaching Excellence with 4.9/5
  • Deliver workshops in both hybrid and in-person formats (Up to 4 hours/times a week)
  • Held office hours and answered questions via online forum and email
• BMAN63012 Interest Rate Derivatives (PGT, 2021-2022)
  • Prepared course materials for the lecturer

Feedback from students: Students really like the clear and step-by-step explanations as well as the added explanations on the board.

Key Achievements

• Fall 2022: AFA PhD Student Travel Grant.
• Fall 2020: Alliance MBS PhD Studentship.
• Summer 2019: Represented the dissertation group to present the work in Santander, London.
• Summer 2018: Excellent Graduation Thesis for Undergraduate.
• Autumn 2017: Third Prize Scholarship for Outstanding Students.

Professional Experience

• Spring 2020 - Autumn 2020: Quant Assistant, Everbright Securities
  • Researched on stock index options volatility strategies (including applying Gurobi as the optimisation tool).
  • Backtested R-Breaker strategy in the Chinese index futures market and constructed an automatisation connection between the strategy and the programming trading system.
  • Built sentiment factors in the Chinese stock market by applying BERT from Google.
  • Implement Factor Mimicking Portfolio from BlackRock on economic growing risk and inflation risk.
• Spring 2018: Intern, East Area Management Dept., Ping An Medical Technology of Ping An Insurance Company of China, Head Office
  • Organised and updated the information of business opportunities and projects from eight branch offices.
• Spring 2017: Intern, Risk Management Dept., Zhejiang Chouzhou Commercial Bank, Shanghai Branch
  • Data update in the post loan management system.
  • In charge of the routine examinations and approvals of loans.

Skills

• Python, Matlab, Stata, R, C++, HTML, Latex, SPSS, Eviews
• Machine learning, Natural language processing, Deep learning

Languages

• English (Advanced), Mandarin (Native), German (Beginner)

Other Studies

• Built up a web portal for labelling text data by Python and Flask.
• Gave internal presentations and held workshops to explain machine learning and neural network to colleagues from different background.
• Master Dissertation: Implemented artificial neural network (FNN, RNN and LSTM) to predict stock realised volatility.
• Built up a multinomial logistic model to predict issues of seasoned equity, convertible bond, and straight bond. Gained an out-of-sample accuracy of 71.0%.
• Used gaussian process to identify different subjects in a dataset measuring people’s movement. chose a feature to simulate traces and achieved average out-of-sample accuracy of 97.65%.
• Applied AIC and BIC for model assessment, forward and backward stepwise as well as Ridge and LASSO regularisation for feature selection.
• Implemented principal component analysis (PCA) and factor analysis (FA) on an anthropometric dataset.
• Applied k-nearest neighbour (KNN) and RBF kernel SVM to classify a biomechanical features dataset.
• Used customised VGG net (including dropout and batch normalisation) to classify CIFAR-10 dataset and gained a 10-time mean accuracy of 80.69% for the test set.
• Used Markov chain Monte Carlo (MCMC) to decrypt a cypher text.
• Used binomial trees, Monte-Carlo method (with variance reduction techniques), finite difference methods and Longstaff-Schwartz least-squares approach to simulate option prices by Excel and C++.
• Simulated stock price paths under the physical and the risk-neutral probability measure to price derivatives and compared the results with those produced by the Black-Scholes formula.
• Programmed by C++ and Matlab to discover some strategies for carpark revenue management.
• Calibrated the Hull-White one factor model to the caplet prices.
• Applied Z-Score, O-Score and KMV-Merton in Matlab to estimate company’s default probability.

Extracurricular Activities

• Autumn 2016 - Summer 2018: Lab Assistant
• Autumn 2015: Assistant Instructor for freshmen
• Autumn 2014 - Summer 2015: Cadre of Sports Department