Of course only the performance of the ultimate 6th place team matters, but conference performance has a big effect on that. As an example, last year the East as a whole had 462 points. This year if the PPG projections hold up, it will have only 425 points. That’s a variance of more more than 30 points which averages out to 3 points per spot. Of course it’s not evenly distributed. If there were a bigger sample I expect we would see a bell curve, but with 10 teams we probably get a bell curve with a lot of random chance mixed in. But a 35-40 point increase or decrease cannot help but have an effect on 6th place, unless all of the extra or missing points cluster very specifically.And generally a sudden spurt or fall off by the sixth place team self-corrects, because it will become the 5th or 7th place team. Right now the end of year distance between 1st and last projects to be about 24 points. Last year it was 30. Either way it is roughly an average 3-point gap between each spot. So on average once a team gets 3 points better or worse than previously projected it moves to a new spot and the team it flips with sets the new standard for sixth place. Unless there is an existing large gap specifically between spots 5-6-7 then individual team performance is undone by this effect.
Last year the East did have a decently sized 5 point gap between 6th and seventh. So TFC who finished 6th could have turned 1 win and 1 tie each into a loss, dropped 4 points, and still finished ahead of Orlando. So TFC’s specific performance did matter to that extent and the playoff line could have been 45 with nothing else changing. Then OSC would be kicking themselves about every single late bad result they got last year to drop points. In the other direction, however, if TFC had won 4 more points the playoff line would only move up 1, because New England sat above them at 50. And if Toronto got one of it’s extra points at the expense of New England, effectively taking 2 from the Revs, than NER would be down 2 and the line would move 1-point to 48. That’s how individual team performance self-corrects and the total points won by the conference becomes more determinative.
In the end I would guesstimate the effect is 60/40 in favor of the conference as a whole, and maybe 50/50. But rarely can I see an individual team’s performance making a difference of more than 4-5 points, and then only likely in one direction like TFC last year.
Finally, the only trend we can even try to predict is the overall conference performance, because there is no way to forecast that Montreal will win 6 of its last 8 or we would drop 6 of the last 9 like happened last year.* So I think the thing to do is predict the line based on the conference as a whole but allow for a few point variance or so either way at the end based on a couple of key teams having a good or bad streak at the end.
* As bad as we were, that was below our season average.
I will always try to be the guy — especially in sports — who says streaks are random, and not due to coaching, new players, or clutch performance.
I could write pages on this but I’ll just state my premises and conclusions: As a species we excel at noticing patterns and coming up with explanations for them. It’s how we survived, prospered and developed. Smart people are good at it and dumb people are good at it. The explanations of smart people tend to make more sense and be right more often, but every one of us including the idiots can notice patterns and come up with plausible-sounding theories in our sleep. Otherwise we would lack both high technology and popular conspiracy theories. It’s in our nature.
But any statistician on the planet will tell you that in any true random distribution you will find streaks, clusters and patterns that have no meaning or rationale whatsoever. And as a species we suck at admitting this in practical applications. We especially suck at it in sports, because we want our heroes to be clutch and our goats to have flaws. But much, much more often than we are willing to admit, the hot streaks, slumps, and performance in big moments just reflect random distributions.