> Should a malicious actor try to tamper with the underlying protocol by using a large number of validators to revert a finalized block (the equivalent of a "51% attack" in PoW) their funds are slashed

But can't the malicious actor estimate very well whether their attack will succeed beforehand?