Notes
Notes
Short pieces that are not full case studies: something learned mid-analysis, a technique worth writing down, a chart that needed a comment. Cases take weeks; these do not.
My risk bands changed every time I rebuilt the database
The query was correct and every number it produced was plausible. Half the rows still changed between two runs of the same pipeline over the same data — because of a tie-break I had not written.
The obvious filter would have deleted what I was measuring
Keeping only the transfers where both clubs were identified was 93% of the money and looked like cheap hygiene. The rows it dropped were the phenomenon under study.
My analysis was reproducible. My file paths were not.
The scripts were correct and the numbers checked out, but every one of them had the path of the machine it was written on baked in — so nobody else could run a line.