Welcome! I am a PhD candidate in Finance at the School of Business and Economics, Brandeis University. My research spans empirical asset pricing, machine learning and AI in finance, law and finance, and market microstructure.
I will be on the job market in the 2026-2027 academic year.
Email: ketianguan@brandeis.edu
Address: 415 South St, Waltham, MA 02453, USA
References: Sida Li (Chair), Anna Scherbina and Yang Sun
• The Nonlinear Ex-Dividend Day: How Dividend Size Reconciles Taxes and Frictions
◦ In a perfect market, a stock's price should fall by exactly the dividend amount on ex-dividend days, but in practice it falls by 85 cents per dollar. In this paper, I argue that taxes explain the puzzle for large dividends, but trading frictions are the dominant force for small dividends. I show substantial heterogeneity in this adjustment: the price-drop ratio is close to 95 percent for large dividends but only about 40 percent for small dividends. This paper is the first to document a systematic nonlinear relationship between dividend size and the ex-dividend price-drop ratio. Using the randomized U.S. Tick Size Pilot as an exogenous coarsening of the price grid, I show that ex-dividend pricing responds to microstructure frictions, with effects concentrated in small-dividend events. Evidence from the 2003 dividend tax cut further supports the tax channel. These results reconcile the two classical explanations. For large dividends, the implied marginal tax rate is approximately 4.7 percent, far below the 15 percent embedded in the conventional 0.85 benchmark. The nonlinear pricing mechanism implies economically significant overnight alpha around large-dividend events for tax-advantaged investors.
◦ Award: Runner-Up Paper Award (GLOBAFA, 2026)
◦ Presentations: FMA Doctoral Student Consortium (scheduled), SFA (scheduled), FIU-ASFAAG Conference, 6th Finance and Accounting Annual Research Symposium, 42nd AFFI, 9th Dauphine Finance PhD workshop, IRMC (2026), MRS International Risk Conference (2026) and GLOBAFA (2026)
◦ Previously circulated as "Nonlinear Ex-Dividend Pricing: Disentangling Taxes and Frictions"
• Do Exogenous Uninformed Order Flows Move Stock Prices? with Sida Li
Revise and Resubmit, Review of Asset Pricing Studies
◦ The literature suggests that stock prices can be influenced by exogenous order flows, even when they do not convey any information about future cash flows. Empirical studies employ various identification strategies to test this hypothesis, though it is difficult to find an exogenous, unexpected large order flow uncorrelated with cash flow news. In this paper, we analyze a large, exogenous, unprecedented asset purchase program around the boundaries of the CSI 500 and CSI 1000 indices. These boundaries are predetermined by market capitalization rankings well in advance of the asset purchase program. Stocks in the CSI 500 index receive a significant exogenous purchase equivalent to 4.49% of their market capitalization, while stocks in the CSI 1000 index receive only 0.51%. We find that CSI 500 stocks earn a 6.4% higher Fama-French 5-factor alpha.
◦ Presentations: FMA (2025), SFA (2025), 41st AFFI, Boston Area Finance Symposium (2025)*, Finance and Accounting Annual Research Symposium (2025), FMARC (2025), Research Symposium on Finance and Economics (2025), 33rd Annual Conference on Pacific Basin Finance, Economics, Accounting, and Management, ISAFE (2025), International Doctoral Forum at CUFE (2025), MRS International Risk Conference (2025), World Finance Conference (2025), 2nd Modern Finance Conference (2025), GLOBAFA Conference (2025), RCEA (2025) and 32nd Annual Global Finance Conference*
• Friends Seeking Influence: Identifying Commonalities in Legal Risk with Aparna Mathur and Anna Scherbina
Invited for submission to the Review of Asset Pricing Studies (dual submission)
◦ We find that Supreme Court decisions have broad implications for U.S. firms, affecting market values not only of companies directly involved in a case but also of other firms exposed to similar legal issues. We identify these indirectly affected firms as publicly traded companies that submit amicus curiae (friend-of-the-court) briefs to influence the Court by advancing business-related arguments in support of their preferred outcomes. Many amici report related litigation currently pending in lower courts and underscore that the Supreme Court’s decision will have substantial implications for their businesses. We find that amici experience no significant stock-price reaction when the side they support wins but earn significantly negative abnormal returns when their side loses. Moreover, a higher fraction of publicly traded amici supporting a particular side of a case significantly increases the likelihood of that side winning. These findings have implications for potential conflicts of interest among Supreme Court justices.
◦ Presentations: FMA (scheduled), SFA (2026, scheduled), Western Economic Association International (WEAI) annual conference*, GLOBAFA, Asian FA (2026) and Brandeis University Brown bag*
• Understanding the Convolutional Neural Network’s Stock Return Predictability with Sida Li and Jia Si
◦ Jiang, Kelly, and Xiu (2023) demonstrate that a convolutional neural network (CNN) trained on stock price chart images achieves striking out-of-sample return predictability. Building on their framework, this paper aims to understand the origin of the CNN’s predictive power. We investigate which visual elements drive the CNN’s predictions, whether richer image content can improve them, and where the model focuses its attention within the image. Controlled ablation experiments establish a clear hierarchy among image elements: OHLC price geometry is the dominant source of predictive power, volume bars contribute marginally, and richer image content uniformly degrades performance—a “less is more” result. Saliency analysis reveals a novel attention–performance paradox: broader, more diffuse scanning of the OHLC geometry yields superior out-of-sample predictions. A baseline long–short strategy achieves an annualized Sharpe ratio of 3.7 over 2020–2024, confirming that the predictability remains robust beyond the original sample period.
◦ Presentations: 3rd AI in Finance Conference (scheduled), FMA poster session (scheduled), Selected for the AFA 2027 Special Project’s “Selected Collection of AI-Authored Research Papers” and Brandeis University Brown bag*
* Presented by coauthor