Project TD Tuning — Version 11

# V11 Context V11 didn’t change the model. It changed the rules. We’d been stopping at Week 12 every time — not because the model was broken, but because a threshold we set early in the experiment was ending runs that still had something to say. We adjusted it, and the model kept going. All the way through the season. Into the playoffs. Through the Super Bowl. And then into the following year before the rolling average finally pulled the plug. 288 games. The longest run we’ve had. The wall at Weeks 11 and 12 is still there in the numbers — you can see it — but it wasn’t fatal. The model recovered and finished strong. What we learned: the threshold was the problem, not the model. V11 is the same predictions as V10, just allowed to run. That’s worth knowing.