This README file provides information about the data and the computer code used to generate the results presented in Ardia et al. (2025), Examining High-Frequency Patterns in Robinhood Users’ Trading Behavior.

By using the code, you agree to the following rules:

• You must cite the paper in working papers and published papers that use the code.
• You must place the DOI of this data/code repository in a footnote to help others find it.
• You assume all risk for the use of the code.

All datasets are proprietary. We provide pseudo data (a sub-sample of our original dataset made up of 100 stocks whose names have been anonymized) to illustrate the code usage. Since it is pseudo data, the outputs generated by running this code do NOT correspond to the results in the paper.

The computer code is written in R and allows for the replication of all tables and figures in the paper.

The file _install_packages.R will install all dependencies (latest versions) and should be run once before running other programs.

The file _load_packages.R will load all relevant R packages. It is used by the various scripts below.

The folder main contains the core of the code. The file 0_run_all.R is a wrapper calling the files below, sequentially, to generate all required tables and figures (alternatively, the user can run the files below individually).

• file prefixed 1_: generates summary statistics (Tables 1 & 2).
• file prefixed 2_: estimates main regressions (Equation (4)). Please use DailyFreq = TRUE or FALSE at the beginning of the script to generate daily or high-frequency results.
• file prefixed 3_: outputs regression results for Table 3 & Figure 1 (HF) and Table 4 & Figure 2 (Daily). Please use DailyFreq = TRUE or FALSE at the beginning of the script to generate daily or high-frequency results.
• file prefixed 4_: estimates subgroup regressions (Equation (5)). Please use OVvsINT, COVID, and MKTCAP options at the beginning of the script to generate specific results.
• file prefixed 5_: performs hypothesis testing on the three identified behaviors. Please use OVvsINT, COVID, and MKTCAP options at the beginning of the script to generate specific results.
• file prefixed 6_: generates Tables 5–7 and Figures 3–5 from regression and hypothesis test results. Please use OVvsINT, COVID, and MKTCAP options at the beginning of the script to generate specific results.
• file f: contains functions used to perform the analyses and generate the tables and figures.

• db_mktcap.rds contains info on size classifications (SMALL, MID, LARGE).
• DT4Reg.rds is our main high-frequency sample, pre-processed, anonymized and reduced to 100 assets, used to generate all tables and figures in the paper.
• DT4Reg_daily.rds is our main daily sample, pre-processed, anonymized and reduced to 100 assets, used to generate all tables and figures in the paper.

The folder outputs is populated by the various tables and figures (and also various .rds files containing the regression results) generated when running scripts in folder code/main. The naming and numbering match the one in the paper.

As mentioned above, since it is pseudo data, the outputs generated by running this code do NOT correspond to the results in the paper.

Ardia D., Aymard C., and Cenesizoglu T. 2025. Examining High-Frequency Patterns in Robinhood Users’ Trading Behavior. International Review of Financial Analysis. https://doi.org/10.1016/j.irfa.2025.104369.