With my NFL experience of how bad coaches tend to be at their jobs, it's likely this experiment worked so well that it will never ever be tried again...
On a more serious note, why do you think it is that lines tend to trend worse once they get more time together? I suppose regression to mean is acceptable as an answer, but if this is a line in the performance bracket you were discussing, in theory composed of three really good players, regression towards 50% shouldn't be expected right?
Perhaps this is my innocence shining through. What is a really good shot%? 55? 60? I don't look over line numbers enough to know the answer to this question. Also, do you have any idea how long lines need to be together for their numbers to begin to stabilise? Surely just 100 minutes falls short, but even if it does are there any numbers that tend to remain consistent even in small samples?
Yeah, some regression to the mean, mostly. Even nearly 2 hours of ice time is still enough for, say, easy competition to have a pretty big impact on the results. I’d say 55%-ish is what should be expected from a good top line. 60% is still rolling over the competition.
I’m not sure if anyone has looked at how long it takes for a line’s results to stabilize. At a guess, I’d say 20-25 games, about double what the Johnston line got, because I think that’s when a team’s overall numbers typically stabilize.
With my NFL experience of how bad coaches tend to be at their jobs, it's likely this experiment worked so well that it will never ever be tried again...
On a more serious note, why do you think it is that lines tend to trend worse once they get more time together? I suppose regression to mean is acceptable as an answer, but if this is a line in the performance bracket you were discussing, in theory composed of three really good players, regression towards 50% shouldn't be expected right?
Perhaps this is my innocence shining through. What is a really good shot%? 55? 60? I don't look over line numbers enough to know the answer to this question. Also, do you have any idea how long lines need to be together for their numbers to begin to stabilise? Surely just 100 minutes falls short, but even if it does are there any numbers that tend to remain consistent even in small samples?
Yeah, some regression to the mean, mostly. Even nearly 2 hours of ice time is still enough for, say, easy competition to have a pretty big impact on the results. I’d say 55%-ish is what should be expected from a good top line. 60% is still rolling over the competition.
I’m not sure if anyone has looked at how long it takes for a line’s results to stabilize. At a guess, I’d say 20-25 games, about double what the Johnston line got, because I think that’s when a team’s overall numbers typically stabilize.