A Machine Learning Perspective on using Accounting data to forecast GDP growth

Top -10 download in wealth, risk mgmt., macro, accounting on ssrn

In joint work with HBS faculty, we find that accounting variables relating to profits, accrual estimates, capital raises, and capital allocation decisions are most informative for the longer-term outlook for the economy. We also verify that non-linear techniques like Random Forest can help considerably.

Privacy vs. Alpha

Is there really a trade-off?

Do we really need to sacrifice privacy to make better investment decisions? Or maybe too much personal information is not only socially risky but also simply adds trading noise! Joint work with Neil Seeman , the Chairman of leading internet-based global trend tracker (RIWI).

Macro Prediction Book chapter

Pre-print copy. Final in the Handbook of US Consumer, Elsevier 2019.

Traditional macroeconomic data used by economic agents to make decisions are noisy, lack richness, and produced with considerable lag. This chapter explores how alternative, web-scale data sources (“Big Data”) can help. We present a case study using a common alternative data source- web search to predict one of the most important data releases- non-farm payrolls (NFP). We discuss the efficacy of various machine learning (ML) techniques, the live performance of alternative data prediction models and the typical problems faced in practice.

The Psychology of Job Loss

ICWSM paper using twitter data

Losing our jobs can make us angry or sad, but can people being angry or sad predict unemployment? We use 1.2B tweets to find out. ICWSM 2016. Joint work with Scott Counts at Microsoft Research and Davide Proserpio at USC.

Predicting Gamestop Reddit Saga?

ICWSM 2017 paper

Quality of information in echo chambers is not very high! Save your money by listening to others who disagree with you. Joint with Scott Counts at Microsoft Research, Ben Zhao and his amazing students.

Improvised Marketing Interventions in Social Media.

Best paper award (Aug 2021), Journal of Marketing.

Does quick marketing wit on social media matter for firm value? We investigate the online phenomenon of improvised marketing interventions (IMI) - social media actions that are composed and executed in real time proximal to an external event, using five multimethod studies and point to features that help reap online sharing and firm value benefits.