When the initial data were released in 2016, we talked about how the market was overreacting on the downside.
At that time, I think people were looking at every incremental IO target as being on the order of magnitude of PD-1. Since then, there has been IDO-like "lackluster" data from a couple of the newer ones. I think people are recalibrating their expectations, with execution and time to market (rather than blowout phase 1/2 data) becoming key.