Quote:
Originally Posted by rubecube
Ah ok, I see. I debated doing a range of outcomes for the teams instead of total W/L records but thought that might be too confusing. I'm doing this more as a relationship to the betting market than I am an overall prediction of how the league will turn out this year. You'll notice I also have 7/8 of 2018's division winners repeating, but we know from history that the likelihood of that happening is very slim.
My hope down the line if I can keep track of this for several seasons is to have enough of a database to determine which stats should be weighted more heavily than others. Right now I weigh PEW differential and record in one-score games the heaviest. I think there likely needs to be other stats included such as third down conversion rate, redzone success, etc. Anything that's high variance from year to year and not necessarily based on skill.
|
Yeah, there is a difference between projections vs. predictions.
Than again, history might show that it's highly unlikely that 7 of 8 teams that win Divisions will happen again, but it does not mean it won't.
So as you say over time when you get a bit more data established as benchmarks of the past, you might be able to see which factors should be weighted to more accurately project outcomes.
It's an interesting analysis, so I'll certainly watch with a bit of interest to see how it plays out.
Realistically if you project within a win on 18 of these you did bloody well.
My overall observation here is that no one team projects to be very dominating, and few teams projected to be bad. So it really does show how the more fluctuating factors can really move the needle for teams being good or bad.