The OP article found "that managers who outperform on bad days do not outperform on other days, while those who outperform on other days tend to underperform on bad days... For past bad-day outperformers, monthly alpha is driven entirely by outperformance on the single worst day of the month. In contrast, past other-day outperformers generate alpha only on normal days and strictly underperform on bad days."

There are two non-overlapping groups: bad-day outperformers and other-day outperformers.

The article rejects some theories: "We reject explanations for persistence related to flow-induced selling pressure or stale pricing and conclude that outperforming on bad days requires skill. What is the nature of this skill? One possibility is that managers who outperform on bad days are simply more skilled overall. The stronger response of flows to performance on bad days could even suggest that such performance provides a better signal of general skill than performance on other days. Both interpretations are unlikely."