Abstract: We let “Algorithmic Market-Makers” (AMMs), using Q-learning algorithms, choose prices for a risky asset when their clients are privately informed about the asset payoff. We find that AMMs learn to cope with adverse selection and to update their prices after observing trades, as predicted by economic theory. However, in contrast to theory, AMMs charge a mark-up over the competitive price, which declines with the number of AMMs. Interestingly, markups tend to decrease with AMMs’ exposure to adverse selection. Accordingly, the sensitivity of quotes to trades is stronger than that predicted by theory and AMMs’ quotes become less competitive over time as asymmetric information declines.
“The horizon of investors’ information and corporate investment” with Olivier Dessaint and Laurent Frésard. November 2022.
Abstract: We study how the quality ofinvestors’information across horizons influences investment. In our theory, managers care about how investment is impounded in current stock prices. Because prices imperfectly reflect investment’s value, they under-invest. However, they under-invest less when investors have better information about thehorizon matching that of their projects. Using a measure of projects’ horizon obtained from the text of regulatory filings, we find that improvements in investors’ long-term (short-term) information induce firms with long-term (short-term) projects to invest more, especially when managers focus on current stock prices. Therefore, the quality ofinvestors’information across horizons has real effects.
“Equilibrium Data Mining and Data Abundance”, with Jérôme Dugast. Revised March 2023.
Abstract: We study, using a noisy rational expectations framework, how the availability of new data to forecast asset payoffs (“data abundance”) affect the capital allocated to quantitative asset managers (“data miners”) relative to other active asset managers, the mean and the cross-sectional dispersion of their performance, and price informativeness. Data miners search for predictors of asset payoffs and trade when they find one with a sufficiently high precision. Data abundance raises the precision of the best predictors. Yet, it eventually induces data miners to lower the bar for their signal precision. Then, their performance becomes more dispersed, and they receive less capital. Overall, data abundance is both a catalyst and an impediment to the rise of quant funds.
“Does Alternative Data Improve Financial Forecasting? The horizon effect”, with Olivier Dessaint and Laurent Frésard. Revised June 2022. Forthcoming Journal of Finance.
Latest draft (SSRN):Here; Online Appendix: Here; Slides: Here
Abstract: Existing research suggests that alternative data is mainly informative about short-term future outcomes. We show theoretically that the availability of short-term oriented data can induce forecasters to optimally shift their attention from the long-term to the short-term because it reduces the cost of obtaining short-term information. Consequently, the informativeness of their long-term forecasts decreases, even though the informativeness of their short-term forecasts increases. We test and confirm this prediction by considering how the informativeness of equity analysts’ forecasts at various horizons varies over the long run and with their exposure to social media data.
Old working papers
“Linkage Principle, Multidimensional Signals and Blind Auctions”, with Stefano Lovo, 2004 (draft on SSRN)
“Price formation and order placement strategies in a dynamic order driven markets”, 1995 (draft)
Thierry Foucault 