×
Rock Chalk Talk: Basketball
Anything pertaining to basketball: college, pro, HS, recruiting, TV coverage
Anything pertaining to basketball: college, pro, HS, recruiting, TV coverage
2019-2020 season projection
Share this page:
4 years 5 months ago #23689
by NotOstertag
"When I was a freshman, I remember Coach Naismith telling us how important it was to play good defense." - Mitch Lightfoot
Poppycock! Balderdash! 
We're going to beat Duke. You'd think that Sagarin (and others) would come up with a more predictive formula to start the season. Duke's current squad might very well be very good, but they're not the team with Zion Williamson, so that alone might be a 5 or 6 point difference. Also, KU isn't the team they were last year with Silvio back, Dok healthy, and some added outside shooting firepower.
You'd think that the ratings guys could apply some kind of formula that could "reset" the rankings until enough games are in the books to make the guesstimates a little more accurate. For example:
>Returning scoring vs. scoring lost. For example, Duke loses Williamson and Barrett. If they amounted to, say, 30% of Duke's scoring how much of that can be replaced with incoming freshmen or former bench guys getting promoted. Maybe that still leave them 10% short of where they were offensively. Decrease their rating accordingly.
>Use a "benchmark" score of the "average staring player" in D1. Then do an educated guess on guys who you don't have data on: to 10 recruit: 20% higher than the benchmark. Top 50 recruit: 10% higher, etc.
>Adjustments for end-of-season status vs. beginning-of-new-season status. For example, KU finished last year with Dok and DeSousa out of the lineup. We do have data on them from when they were playing, so we can make decent predictions about their potential being healthy and eligible. Maybe that alone bumps KU's initial rating up a couple points.
In the end, I guess it doesn't matter because once enough games are played, it all balances out. Nevertheless, I'd love it if some of these ratings experts did a little work to handicap the teams based on what we know going in.
In any case, that's Asteroid for keeping us all on top of this stuff.
We're going to beat Duke. You'd think that Sagarin (and others) would come up with a more predictive formula to start the season. Duke's current squad might very well be very good, but they're not the team with Zion Williamson, so that alone might be a 5 or 6 point difference. Also, KU isn't the team they were last year with Silvio back, Dok healthy, and some added outside shooting firepower.
You'd think that the ratings guys could apply some kind of formula that could "reset" the rankings until enough games are in the books to make the guesstimates a little more accurate. For example:
>Returning scoring vs. scoring lost. For example, Duke loses Williamson and Barrett. If they amounted to, say, 30% of Duke's scoring how much of that can be replaced with incoming freshmen or former bench guys getting promoted. Maybe that still leave them 10% short of where they were offensively. Decrease their rating accordingly.
>Use a "benchmark" score of the "average staring player" in D1. Then do an educated guess on guys who you don't have data on: to 10 recruit: 20% higher than the benchmark. Top 50 recruit: 10% higher, etc.
>Adjustments for end-of-season status vs. beginning-of-new-season status. For example, KU finished last year with Dok and DeSousa out of the lineup. We do have data on them from when they were playing, so we can make decent predictions about their potential being healthy and eligible. Maybe that alone bumps KU's initial rating up a couple points.
In the end, I guess it doesn't matter because once enough games are played, it all balances out. Nevertheless, I'd love it if some of these ratings experts did a little work to handicap the teams based on what we know going in.
In any case, that's Asteroid for keeping us all on top of this stuff.
"When I was a freshman, I remember Coach Naismith telling us how important it was to play good defense." - Mitch Lightfoot
Please Log in or Create an account to join the conversation.