Subject: Re: How long before a published screen “stops working”
Academic evidence: Across 97 anomalies the excess return falls by ≈ 32 % after publication as capital crowds in. It does not go to zero.
SSRN

Do you have a link to the paper?

DB2