Research

Successful quantitative investment usually relies on precise predictions of the future movement of the stock price. Recently, machine learning based solutions have shown their capacity to give more accurate stock prediction and become indispensable components in modern quantitative investment systems. However, the i.i.d. assumption behind existing methods is inconsistent with the existence of diverse trading patterns in the stock market, which inevitably limits their ability to achieve better stock prediction performance. In this paper, we propose a novel architecture, Temporal Routing Adaptor (TRA), to empower existing stock prediction models with the ability to model multiple stock trading patterns. Essentially, TRA is a lightweight module that consists of a set of independent predictors for learning multiple patterns as well as a router to dispatch samples to different predictors. Nevertheless, the lack of explicit pattern identifiers makes it quite challenging to train an effective TRA-based model. To tackle this challenge, we further design a learning algorithm based on Optimal Transport (OT) to obtain the optimal sample to predictor assignment and effectively optimize the router with such assignment through an auxiliary loss term. Experiments on the real-world stock ranking task show that compared to the state-of-the-art baselines, e.g., Attention LSTM and Transformer, the proposed method can improve information coefficient (IC) from 0.053 to 0.059 and 0.051 to 0.056 respectively. Our dataset and code used in this work are publicly available.

Modeling and managing portfolio risk is one of the most important step to achieve growing and preserving investment performance. Within the modern portfolio construction framework that built on Markowitz's theory, the covariance matrix of stock returns is required to model the portfolio risk. Traditional approaches to estimate the covariance matrix are based on human designed risk factors, which often requires tremendous time and effort to design better risk factors to improve the covariance estimation. In this work, we formulate the quest of mining risk factors as a learning problem and propose a deep learning solution to effectively "design" risk factors with neural networks. The learning objective is carefully set to ensure the learned risk factors are effective in explaining stock returns as well as have desired orthogonality and stability. Our experiments on the stock market data demonstrate the effectiveness of the proposed method: our method can obtain 1.9% higher explained variance measured by R² and also reduce the risk of a global minimum variance portfolio. Incremental analysis further supports our design of both the architecture and the learning objective.

Targeting, as a widely deployed personalization strategy that takes marketing actions in a specific group of customers (e.g, advertising, promotion, pricing), requires companies to continually run randomized experiments. In a customer changing environment, individual responsiveness to the targeting treatments shifts, known as a problem called concept drift. Such a shift makes two things challenging: how to collect new experimental data with a limit budget, and how to utilize historical data where not all the data can be accessed from time to time. In this paper, we address these by proposing an active sampling method to prioritize sampling on changing customers, and a selective memory mechanism to memorize customers in the buffer. We demonstrate the superiority of our method through simulation and a real-world application.

Understanding and predicting consumer preferences in dynamic markets is crucial for marketing decision-making. Existing methods for modeling dynamic consumer heterogeneity are either inflexible in representing individual-level preference trajectories or as in the case of recently proposed Gaussian process-based approaches, induce unimodal cross-sectional heterogeneity distributions. This prevents them from accurately capturing the growth, decline, or divergence of consumer segments over time. We introduce a novel Bayesian nonparametric machine learning framework that leverages a dependent Dirichlet process for modeling heterogeneity to overcome these methodological constraints. Our approach combines the strengths of Dirichlet processes and Gaussian processes to flexibly model a collection of temporally correlated population distributions. This structure simultaneously captures smooth individual-level preference trajectories and evolving multimodal population distributions. Extensive simulations show that our model significantly outperforms a state-of-the-art GP-based benchmark in recovering parameters and predicting choices when the evolving population distributions are multimodal and performs comparably when they are normal. An application to IRI scanner data for multiple categories spanning the Great Recession shows that our model provides more plausible price elasticity estimates, improved predictions, and granular insights into preference shifts during the recession. We finally show that coupon targeting strategies based on our model can generate higher profits.

Experience goods such as sporting and artistic events, songs, videos, news stories, podcasts, and television series, are often packaged and consumed in bundles. Many such bundles are ordered in the sense that the individual items are consumed sequentially, one at a time. We examine if an individual's decision to consume the next item in an ordered bundle can be predicted based on his/her consumption pattern for the preceding items. We evaluate several predictive models, including two custom Transformers using decoder-only and encoder-decoder architectures, fine-tuned GPT-3, a custom LSTM model, a reinforcement learning model, two Markov models, and a zero-order model. Using data from Spotify, we find that the custom Transformer with a decoder-only architecture provides the most accurate predictions, both for individual choices and aggregate demand. This model captures a general form of state dependence. Analysis of Transformer attention weights suggests that the consumption of the next item in a bundle is based on approximately equal weighting of all preceding choices. Our results indicate that the Transformer can assist in queuing the next item that an individual is likely to consume from an ordered bundle, predicting the demand for individual items, and personalizing promotions to increase demand.