The mistake John makes in each case is ignoring the regression to the mean. Anytime we have a baseline established and then a certain data point lies away from that baseline, our brains see a pattern. The further away the data point, the more we want to assign a cause. In addition, the further away the data point, the more likely subsequent data points will be closer to the average than to the exception. This fallacy is committed when we assign arbitrary reasons for the return to mediocrity, rather than acknowledging that it is expected.
1) Notice John does not just attempt to explain his exceptional time with wheat grass juice, he is also trying to explain why he did not continue having exceptional performance even though he continues drinking juice. His return to an average time is evidence that the wheat grass did nothing, but rather than abandon his original mistake, he explains it away with a further assumption.
2) If the remedy worked we would expect the pain to vanish, not simply return to a previous average level. The presence of pain indicates the cause has not been treated. It is much more likely that the pain would return to its baseline on its own, in fact, that is what we expect.
3) The more exceptional John and Jane's height, the less likely their children are to equal or exceed that exception. We do expect tall parents to have tall children, but if all children reached the height of their parents then John's height wouldn't be exceptional. The taller the parents are, the more likely their children will be closer to average than to the exception.