I wrote a script to get UTC time for each blockheight and then plotted monthly stale block counts from historical data. Over the last 6 years (2018-2023), we had an average of just 2.08 stale blocks per month observed in this dataset.