The multi-factor model is a widely used model in quantitative investment. The success of a multi-factor model is largely determined by the effectiveness of the alpha factors used in the model. This paper proposes a new evolutionary algorithm called AutoAlpha to automatically generate effective formulaic alphas from massive stock datasets. Specifically, first we discover an inherent pattern of the formulaic alphas and propose a hierarchical structure to quickly locate the promising part of space for search. Then we propose a new Quality Diversity search based on the Principal Component Analysis (PCA-QD) to guide the search away from the well-explored space for more desirable results. Next, we utilize the warm start method and the replacement method to prevent the premature convergence problem. Based on the formulaic alphas we discover, we propose an ensemble learning-to-rank model for generating the portfolio. The backtests in the Chinese stock market and the comparisons with several baselines further demonstrate the effectiveness of AutoAlpha in mining formulaic alphas for quantitative trading.

Empirical data reveals that the liquidity flow into the order book (depositions, cancellations andmarket orders) is influenced by past price changes. In particular, we show that liquidity tends todecrease with the amplitude of past volatility and price trends. Such a feedback mechanism inturn increases the volatility, possibly leading to a liquidity crisis. Accounting for such effects withina stylized order book model, we demonstrate numerically that there exists a second order phasetransition between a stable regime for weak feedback to an unstable regime for strong feedback,in which liquidity crises arise with probability one. We characterize the critical exponents, whichappear to belong to a new universality class. We then propose a simpler model for spread dynamicsthat maps onto a linear Hawkes process which also exhibits liquidity crises. If relevant for thereal markets, such a phase transition scenario requires the system to sit below, but very close tothe instability threshold (self-organised criticality), or else that the feedback intensity is itself timedependent and occasionally visits the unstable region. An alternative scenario is provided by a classof non-linear Hawkes process that show occasional "activated" liquidity crises, without having to bepoised at the edge of instability.

We analyse the optimal exercise of an American call ESO written on a stock whose drift parameter falls to a lower value at a \emph{change point}, an exponentially distributed random time independent of the Brownian motion driving the stock. Two agents, who do not trade the stock, have differing information on the change point, and seek to optimally exercise the option by maximising its discounted payoff under the physical measure. The first agent has full information, and observes the change point. The second agent has partial information and filters the change point from price observations. This scenario is designed to mimic the positions of two employees of varying seniority, a fully informed executive and a partially informed less senior employee, each of whom receives an ESO. The partial information scenario yields a model under the observation filtration $\widehat{\mathbb{F}}$ in which the stock drift becomes a diffusion driven by the innovations process, an $\widehat{\mathbb{F}}$-Brownian motion also driving the stock under $\widehat{\mathbb{F}}$, and the partial information optimal stopping value function has two spatial dimensions. We rigorously characterise the free boundary PDEs for both agents, establish shape and regularity properties of the associated optimal exercise boundaries, and prove the smooth pasting property in both information scenarios, exploiting some stochastic flow ideas to do so in the partial information case. We develop finite difference algorithms to numerically solve both agents' exercise and valuation problems and illustrate that the additional information of the fully informed agent can result in exercise patterns which exploit the information on the change point, lending credence to empirical studies which suggest that privileged information of bad news is a factor leading to early exercise of ESOs prior to poor stock price performance.

Individuals' output often depends not just on their ability and actions, but also on external factors or fundamentals, whose effect they cannot separately identify. At the same time, many individuals have incorrect beliefs about their own ability. Heidhues et al. (2018) characterise overconfident and underconfident individuals' equilibrium beliefs and learning process in these situations. They argue overconfident individuals will act sub-optimally because of how they learn. We carry out the first experimental test of their theory. Subjects take incorrectly marked tests, and we measure how they learn about the marker's accuracy over time. We use machine learning to identify heterogeneous effects. Overconfident subjects have lower beliefs about the fundamental, as Heidhues et al. predict, and thus would make sub-optimal decisions. But we find no evidence it is because of how they learn.

The VSTOXX index tracks the expected 30-day volatility of the EURO STOXX 50 equity index. Futures on the VSTOXX index can, therefore, be used to hedge against economic uncertainty. We investigate the effect of trader inventory on the price of VSTOXX futures through a combination of stochastic processes and machine learning methods. We formulate a simple and efficient pricing methodology for VSTOXX futures, which assumes a Heston-type stochastic process for the underlying EURO STOXX 50 market. Under these dynamics, approximate analytical formulas for the implied volatility smile and the VSTOXX index have recently been derived. We use the EURO STOXX 50 option implied volatilities and the VSTOXX index value to estimate the parameters of this Heston model. Following the calibration, we calculate theoretical VSTOXX future prices and compare them to the actual market prices. While theoretical and market prices are usually in line, we also observe time periods, during which the market price does not agree with our Heston model. We collect a variety of market features that could potentially explain the price deviations and calibrate two machine learning models to the price difference: a regularized linear model and a random forest. We find that both models indicate a strong influence of accumulated trader positions on the VSTOXX futures price.

Public transit disruption is becoming more common across different transit services, which can have a destructive influence on the resiliency and reliability of the transportation system. Utilizing a recently collected data of transit users in the Chicago Metropolitan Area, the current study aims to analyze how transit users respond to unplanned service disruption and disclose the factors that affect their behavior.

We prove the existence of an equilibrium in a model with transaction costs and price impact where two agents are incentivized to trade towards a target. The two types of frictions -- price impact and transaction costs -- lead the agents to two distinct changes in their optimal investment approach: price impact causes agents to continuously trade in smaller amounts, while transaction costs cause the agents to cease trading before the end of the trading period. As the agents lose wealth because of transaction costs, the exchange makes a profit. We prove the existence of a strictly positive optimal transaction cost from the exchange's perspective.

The speeches stated by influential politicians can have a decisive impact on the future of a country. In particular, the economic content of such speeches affects the economy of countries and their financial markets. For this reason, we examine a novel dataset containing the economic content of 951 speeches stated by 45 US Presidents from George Washington (April 1789) to Donald Trump (February 2017). In doing so, we use an economic glossary carried out by means of text mining techniques. The goal of our study is to examine the structure of significant interconnections within a network obtained from the economic content of presidential speeches. In such a network, nodes are represented by talks and links by values of cosine similarity, the latter computed using the occurrences of the economic terms in the speeches. The resulting network displays a peculiar structure made up of a core (i.e. a set of highly central and densely connected nodes) and a periphery (i.e. a set of non-central and sparsely connected nodes). The presence of different economic dictionaries employed by the Presidents characterize the core-periphery structure. The Presidents' talks belonging to the network's core share the usage of generic (non-technical) economic locutions like "interest" or "trade". While the use of more technical and less frequent terms characterizes the periphery (e.g. "yield" ). Furthermore, the speeches close in time share a common economic dictionary. These results together with the economics glossary usages during the US periods of boom and crisis provide unique insights on the economic content relationships among Presidents' speeches.