Assaying Anomalies

This page contains a form for the web application that allows users to test the robustness of a new predictive signal for the cross-section of stock returns by uploading a .csv file with three columns: firm identifier, date, and signal. The application then automatically generates a self-contained report testing the new anomaly as described in Novy-Marx and Velikov (2023), and emails the submitter latex files and .pdf documents for this report.

The automation of this direct submission to the backend server is still under development. We will, however, manually forward signals submitted through the web-based form to the server that runs the publicly available code. Due to computational requirements we can currently only run one signal at a time and, depending on the sample length, each signal can take multiple hours. To discourage excessive usage at this point, we ask that users submit no more than one file per day.

Submission of a file implies that the user has read and agrees to the following disclosure:

Disclosure: Any data, information, analysis and conclusions drawn there from relating to the Assaying Anomalies Web Application (“Web App”) is intended solely for research, education, and non-commercial purposes. The Web App is provided here without representation or warranty of any kind, including but not limited to merchantability, non-infringement of third party intellectual property rights, or fitness for a particular purpose. The Pennsylvania State University has no obligation to provide updates to the Web App, make the Web App available, or maintain the Web App and depictions thereof on this webpage or in any other form. The Pennsylvania State University, its trustees, officers, employees, and agents shall not be liable for any use, distribution, or reliance upon the Web App for any purpose. If you are interested in licensing the Assaying Anomalies Web Application for commercial purposes, please contact Penn State’s Office of Technology Management at otminfo@psu.edu.

Currently, the file also needs to satisfy the following requirements:

1. The first column should have CRSP permno identifiers and have “permno” as its header.
2. The second column should have dates in YYYYMM format and have “dates” as its header.
3. The third column should have values for the signal.
4. The tests require at least ten years of signal data.
5. The signal is assumed to to be signed to positively predict returns.

An example of a valid input file structure would be the following:

Upon submission of the form, the corresponding author will in most cases receive an email within 48 hours with the zipped output of the protocol. The .zip file includes .tex file for the online appendix, .pdf files of the figures, and the log file from running the MATLAB code. See examples of the compiled output in the Used In page. Be sure to check your Junk folder or add assayinganomalies@gmail.com to your safe sender list. In times of high usage, we may may switch off some of the more computationally-intensive combination strategy constructions for Figure 8.

Submission form