…
Thesis: markets undervalue what a new CEO can do, so early evidence that a new strategy is working predicts 1-3 years of outperformance against peers.
Verdict: not supported. Once survivorship bias is removed, companies that change CEO trail similar companies that kept theirs, by about 5-8% after one year and 9-19% after two. No screen we tried (first-month price move, pay design, the hockey-stick shape, or AI-scored traits of the new CEO in a pre-registered test) picked the winners out of sample.
Data: every US 8-K Item 5.02 CEO change, Jan 2020 - Sep 2026, companies above $1bn including those since delisted; prices from Financial Modeling Prep. Design period 2020-22, hold-out Jan 2023 - Jun 2025. This model holds summary results only; the numbers are inputs you can edit.