Strategies also decay. Every quant fund is constantly generating new algorithms, because the value of old algorithms drops over time as others discover them.
Strategies decay if others discover them. RenTech claims to have a strategy that completely removes a certain arbitrage from prices if executed correctly. It would still show up in volume, but you can't extract the critical parameters from anonymized volume.
This would check out as a winning strategy. If you figure how to remove this arbitrage out before anyone else, then you nobody else can replicate your strategies. Since they were early in the quant game, this seems plausible (though other things mentioned here also seem plausible -- insider trading, lots of brilliant people, etc).
surprising how after 30 years no one else has figured out how to to this