I took the same sample of players I used for my year-to-year study, and I re-ran the same experiment as before, except I used in-season splits instead of back-to-back seasons. Specifically, I used the ratio of team dropbacks in odd-numbered games to dropbacks in even-numbered games to try to predict the change in ACY from even to odd-numbered games.

The regression once again had a very low R^2 but a significant coefficient on expected change in ACY, and that coefficient was 0.779!

So it’s all over the place. I think Chase’s 50% number is probably the best bet, if not simply because it splits the difference between all of the different results we’ve seen, and it’s pretty convenient. I’d be willing to settle on that as the “official” discount rate on team dropbacks vs league dropbacks in the TRY formula.