Sunday, March 28, 2021

COVID Modeling Update: Bayesian Analysis

I often shy away from papers by colleagues/coauthors, trying to maintain some semblance of objectivity.  But this one is too cool to let go, "Bayesian Estimation of Epidemiological Models: Methods, Causality, and Policy Trade-Offs," by Arias, Fernandez-Villaverde, Rubio-Ramırez, and Shin.  Non-linear non-Gaussian state space with time-varying parameters are central. See  

Sunday, March 21, 2021

The Latest in Probability Forecast Evaluation

VERY nice and useful paper in the Proceedings of the National Academy by Tilmann Gneiting et al.: "Stable reliability diagrams for probabilistic classifiers",  Supplement:

(Significantly-revised version of "Evaluating probabilistic classifiers: Reliability diagrams and score decompositions revisited“,

Thursday, March 18, 2021

Machine Learning Panel Data

This looks very cool.  Great presenter and great discussant.  March 22.

 SoFiE Seminar

with Eric Ghysels and Max Farrell


Eric Ghysels (UNC Chapel Hill)


“Machine Learning Panel Data Regressions with an Application to Nowcasting Price Earnings Ratios"


Max Farrell (University of Chicago)


March 22, 2021


11am New York / 8am San Diego / 3pm London / 4pm Paris / 11pm Beijing

Zoom Link:


A link to a video recording will be available here soon after the event.

Sunday, March 14, 2021

The Finance Crowd Analysis Project

I have long been interested in crowdsourcing, from a forecast combination perspective. Fincap, described below, is related but different.  I look forward to seeing and pondering the fincap results.

The following material is adapted from the Fincap project site.  For details, including a really slick 2-minute video intro, see

#fincap is the first crowd-sourced empirical paper in Economics/Finance.

More than 100 research teams (RTs) from around the world will test the same set of hypotheses on the same data. They will work independently and write a short academic paper based on their findings.

These reports will be evaluated by more than 30 distinguished academics whom we refer to as peer evaluators (PEs). Their feedback will be passed on to the RTs so that they can revise their papers. 

The project coordinators will study the #fincap results to learn about the scientific process. They have committed ex-ante to a meta-science analysis which was frozen before any instructions and data were given to the RTs and PEs.

Sunday, March 7, 2021

Network Cluster-Robust Inference

Interesting progression of HAC / cluster-robust inference, from serial correlation, to spatial correlation (e.g., Müller and Watson, 2021, here), and now network-induced correlation. 

Very cool new network result from Michael Leung at USC: Asymptotic independence (in the number of linking steps), the key to robust/clustered inference in network environments, holds iff network clusters have conductance (the ratio of edge boundary size to volume) approaching 0. (Yes, necessary and sufficient!)

Friday, March 5, 2021

2021 SoFiE Machine Learning in Finance and Economics Conference TODAY

Starts in a few hours!  Program looks great. Hard to script a better day.    

Registration and program at :

2021 SoFiE Machine Learning Virtual Conference 

Eric Ghysels, UNC Chapel Hill 
Bryan Kelly, Yale University
Dacheng Xiu, University of Chicago

March 5, 2021 

10:00 AM: Introduction 

Session 1: Chair Eric Ghysels 

10:10 – 10:55 Mispricing and uncertainty in international markets, Mirela Sandulescu and Paul Schneider 

Discussant: Rohit Allena 

11:00 – 11:45 A penalized two-pass regression to predict stock returns with time-varying risk premia, Gaetan Bakalli, Stephane Guerrier and Olivier Scaillet 

Discussant: Paolo Zaffaroni 

Session 2: Chair Bryan Kelly 

1:00 – 1:45 The Knowledge Graph for Macroeconomic Analysis with Alternative Big Data, Yucheng Yang, Yue Pang, Guanhua Huang and Weinan E 

Discussant: Phillipe Goulet Coulombe 

1:50 – 2:35 On the Aggregation of Probability Assessments: Regularized Mixtures of Predictive Densities for Eurozone Inflation and Real Interest Rates, Francis X. Diebold, Minchul Shin, Boyuan Zhang

Discussant: Allan Timmermann 

Session 3: Chair Dacheng Xiu 

2:45 – 3:30 High-Frequency Expectations from Asset Prices: A Machine Learning Approach, Aditya Chaudhry and Sangmin S. Oh 

Discussant: Jonathan Wright 

3:35 – 4:20 High-Dimensional Granger Causality Tests with an Application to VIX and News, Andrii Babii, Eric Ghysels and Jonas Striaukas 

Discussant: Markus Pelger 

4:20 – 4:30 Conclusion