I'm not trying to mischaracterize your response, but for me at least, it seems like you say stats don't matter because you disagree with them, used an ancedote as a counter example (big no-no, at least from a logical standpoint), and then defended that stance by saying 'stats and real life are different things', which is counterintuitive; stats are unarguably a measure of 'real life', without actual real life instances, the stats couldn't exist, by definition. Thats just what stats are.
That's not to say that statistics are definitive argument within themselves, just that saying, 'those are wrong because I disagree with them, and I agree with this thing which I cannot support with meaningful numbers' isn't really a constructive counter.
I think what he was saying is not that stats are irrelevant, but that they don't necesarilly tell the full story.
For example what I've seen from Jeff Green so far (even in his 20+ point games) is he tends to score in streaks. He'll score 13 in 5 minutes, then he'll not score for two quarters, then he'll score another 5 points in the last 8 minutes of the fourth. With the exception of his 40 point that seems to be the trend.
I think this is what he meant when he said that Jeff needs minutes to be really effective. If you play him 15-20 minutes a game then the result is going to be random. You might get the "15 points in 20 minutes" Jeff, or you might get the "5 points in 20 minutes" Jeff. However, if you play him 35 minutes every night you're most likely going to get anywhere from 10-25 points from him every night (and > 18 points more often than not).
This is why looking at his scoring per minutes (in the games where he played less minutes) might not really tell the full story.
Look at guys like Pierce or KG and they tend to spread their scoring more evently, so if you gave either of those guys 20 minutes or 40 minutes their production per-minute would probably not vary much because their level of aggressiveness is generally pretty consistent.
I think the point was a little different actually. You're talking about being streaky, which would make his scoring in small doses very unpredictable. But it wouldn't necessarily change the average number of points per minute. If Green plays 20 minutes per game and scores 5 one night and 20 another, he's averaging 12.5. If he gets 10 one night and 15 the next, he still averages 12.5. The variability is lower but the average is the same. And even if the variability goes down when minutes go up, the average might not change (or could go down).
I think the argument is (or should be) that Green's per-36 numbers are better when he plays more minutes.
Right now the data bear that out. Green averages 21 points per 36 minutes as a starter in 11 games this year. Off the bench, he averages 15 points per 36 minutes. One interpretation is that Green actually becomes a better player when you give him more minutes.
I personally disagree with that interpretation for two reasons. First, when Green starts it's because another starter is out, so his role in the offense will naturally increase. If JG is on the court with KG, he will take fewer shots than when he's out there with Bass or Wilcox - and rightly so. And second, Doc might leave Green (or any player) out there more often when he's hot - in which case the nights on which he gets more minutes will of course be the ones on which he's most effective.
But in neither case does giving Green more minutes *cause* him to score more, per minute. To interpret the numbers that way would confuse correlation with causality.