Data mining
Searching many variables or rules until something looks significant. Test enough hypotheses and chance alone produces impressive results, so how many were tried matters as much as the result.
Where you will meet it
Data mining appears in 2 learning outcomes on the CFA syllabus. A few of them:
- Set out what backtesting an investment strategy aims to achieve
“A backtest is a necessary filter for an idea but not proof: good historical performance may reflect luck, data mining or a regime that has passed.”
- Set out the obstacles an analyst meets when producing forecasts for capital markets
“Data mining and time-period bias create false patterns, and ignoring conditioning information hides how relationships change with the state of the economy.”
Practise it
Practise data mining until it is automatic: questions that explain every wrong option.
Take the free diagnostic →