Intraday Momentum and Returns to Liquidity Providers (Job Market Paper)

This paper documents a structural shift in U.S. equity intraday return dynamics. Before the SEC’s decimalization in 2001, five-minute returns displayed negative autocorrelation and reversal; afterward, they exhibited positive autocorrelation and momentum. Evidence from the 2016 Tick Size Pilot and the Russell 2000 reconstitution links this transition to smaller tick sizes and autocorrelated institutional demand. While such momentum typically implies losses for liquidity providers, both cross-sectional return patterns and time-series autocorrelations reveal a clear term structure: reversal and negative autocorrelation at sub-minute intervals transition to momentum and positive autocorrelation at longer horizons. Thus, trading at higher frequencies and exploiting make-take fees and bid-ask spreads can offset these losses, allowing profitability even when providing liquidity against momentum.

Extrapolative Expectations and Corporate Risk Management 

with Haibo Jiang, Nishad Kapadia, and Yuhang Xing

Aggregate hedging by gold producers more than doubled in the 1990s and then fell by 90% in the 2000s. We find that hedging varies inversely with past gold returns, consistent with extrapolative expectations. Three proxies for expected returns: (i) expectations of future gold prices disclosed in annual reports, (ii) analyst survey forecasts, and (iii) exponentially weighted averages of past returns, are extrapolative and predict hedging. Extrapolation has real costs: extrapolators earn lower future stock returns and exhibit smaller, more cyclical investment and profitability. Alternative explanations based on distress, investments, taxes, or financing constraints do not explain the time series of hedging. 

Intraday Liquidity Commonality

with Seyed Mohammad Kazempour and Yuhang Xing

We study liquidity commonality using high-frequency intraday data and document a sharp rise over the past two decades. This increase is driven mainly by a decline in idiosyncratic liquidity variation rather than stronger systematic components. Supply-side explanations receive limited support, as commonality falls during crises due to spikes in idiosyncratic variation. Demand-side factors are strongly associated with higher commonality, yet they cannot fully account for the magnitude of the increase. Our findings underscore the value of high-frequency analysis and reveal a puzzle that calls for new theoretical explanations of liquidity comovement in modern markets.

Expectations and Option Pricing

This paper examines whether and how stock return expectations affect option pricing. I construct expectations based on machine learning methods to incorporate a rich set of predictors. Options with extreme expectations are more expensive and overpriced, especially options with the lowest expectations. I further show that extreme expectations are associated with high option demand. This might indicate that option prices are affected by the demand related to expectations. Overall, this paper complements the intermediary option pricing by proposing stock return expectation as a potential driver of option demand, and documents the relationship between expectations and option expensiveness.