Get historical dates of earning announcements, EPS estimates, and actual EPS from Yahoo Finance. Run python xearnings_bulk.py to create earnings_dates_all.csv. The program needs to have a recent version of yfinance to be installed. Use the command
pip install -U yfinance to do so. itot_tickers_20260115.txt has the symbols of the 2483 stocks owned by the iShares Core S&P Total US Stock Market ETF (ITOT) as of 2026-01-15. They were extracted from the "Detailed Holdings and Analytics" file at the fund site. It takes about 3s per stock to get data for quarterly earnings from 2002 through 2025.
The script xsummarize_earnings.py reads the CSV data file created by xearnings_bulk.py and summarizes the data by stock, creating earnings_summary.csv, which has the # of earnings dates, the first and last earnings dates, the median, mean, and standard deviation of earnings surprises, the # of positive and negative earnings surprises, the CAGR of EPS, and the median, min, and max # of days between earnings dates. For almost all stocks the median # of days between earnings dates is close to 90, since companies report earnings quarterly in the U.S. In many case the min days between earnings dates is much smaller than this. I don't know if these cases correspond to earnings restatements or data errors.
The script xannual_earnings.py also reads the CSV data file created by xearnings_bulk.py and creates annual_earnings.csv, which summarizes the data by stock and year, showing actual and estimated EPS by quarter. If the # of earnings dates in a year exceeds 4, that could indicate an earnings restatement or data error.
The earnings are split-adjusted. A quirk in the Yahoo Finance data is that reported and estimated EPS are shown to the nearest 0.01. For NVDA, which has split many times, this results in some early earnings numbers being displayed as missing, even though earnings surprises are reported. There are a few stocks, such as DCTH and TNXP, where some reported and estimated EPS are very large negative numbers, which is anomalous.