|
|
Post by bulkey on Nov 17, 2021 13:25:58 GMT -5
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 17, 2021 13:40:21 GMT -5
We've only played one game and we gave up 80 points that game, but things will be more settled in a couple of weeks!! Notre Dame @ #11 is funny, but they might end up a top 25 team? |
|
|
|
Post by doggydaddy on Nov 17, 2021 13:44:51 GMT -5
Massey is hysterical this early in the season.
|
|
|
|
Post by phil on Nov 17, 2021 13:49:30 GMT -5
That does surprise me. I won't pretend to know exactly what the algorithm is, but I paid a lot of attention to movement over time. UConn was picked to be Arkansas by 10.5 and normally, exceeding the predicted margin increases your strength and weakens your opponent. I'm sure Ken Massey would say that's an oversimplification, and there's more to it than that but we did exceed the expected margin by 4 1/2 points and I would not have expected that to result in a significant drop.
If I recall correctly, Arkansas was number 30 before the game and is now 34, so the game hurt them, but I'm speculating that both teams got hurt because someone at UConn's level shouldn't be giving up 80 points.
|
|
|
|
Post by swash on Nov 17, 2021 15:29:39 GMT -5
Massey is hysterical this early in the season. For the first few weeks, the Massey calculation data comes mostly from last season. That means some teams like UCONN and SCar, they're pretty good, but Syracuse or Duke ... Nothing. Come to think of it, I wonder how they handled teams like Duke and Vandy (who cancelled their seasons early due to COVID) with essentially no data from last year. I've been in that boat, and there is no good answer. After a single game, this year's win/loss data suggest every team will be either be undefeated or will go winless - which everyone recognizes as preposterous. But last year doesn't take into account graduations and transfers and recruiting classes and coaching turnover and .... The third option is to include a fudge-factor from "the experts", which is also fraught with fracks. Most try to choose more data from each game ... every play or offense, defense, opponent, etc, use that as a core, but early on, you need a base, so last year's data is very tempting, then throw in some wildcards for injuries and changes, maybe sprinkle in a little from the early season polls. Then as the season progresses, slowly phase out that other stuff and let the team's current production speak for itself. The strength of Massey and other calculated rankings is that they are not subject to bias or desires. The downside is that until there is a significant amount of data in hand, things swing wildly. Anyone really think Texas will be better than Stanford all season? No? But that's "all" of the data we have thus far. |
|
|
|
Post by swash on Nov 17, 2021 15:38:09 GMT -5
That does surprise me. I won't pretend to know exactly what the algorithm is, but I paid a lot of attention to movement over time. UConn was picked to be Arkansas by 10.5 and normally, exceeding the predicted margin increases your strength and weakens your opponent. I'm sure Ken Massey would say that's an oversimplification, and there's more to it than that but we did exceed the expected margin by 4 1/2 points and I would not have expected that to result in a significant drop. If I recall correctly, Arkansas was number 30 before the game and is now 34, so the game hurt them, but I'm speculating that both teams got hurt because someone at UConn's level shouldn't be giving up 80 points. All those above us have more data points. I suspect Massey has a "confidence score" included in their secret sauce. One game will be less informative than several, so confidence will be higher. I will also bet that Massey takes offense and defense measures. They consider Ark a bit better than average on Offense and weaker than average on D. So UCONN should have scored a lot on them, but shouldn't have given up so many points. If the team gives up 80 ... or even 72 on average, then we're more likely to see them lose some games. Once they show that they can defend and have more data points to measure, the rating will jump back up |
|
|
|
Post by denverernie on Nov 17, 2021 23:53:06 GMT -5
When you look at projected wins UCONN is best in their rankings but not credited.
|
|
