but your own data shows that the break point -- when GS is likely to fall -- is on the other side of 80 minutes.
estimating, if peak minutes (around 60) produces what looks to be 60% likelihood of a GS increase of 1, then at 70 - 75 minutes, the likelihood (only) drops to around 40%. it is still positive (likely) to produce an increase, and two weeks running -- i have no skills at statistics -- but wouldn't that make very likely to produce an increase?
i will grant the point that what happened to my player is statistically improbable, minutely possible. in other words, bad luck.