UConn is #1 again and…computers don’t like Iowa State

[HuskyNan]

That computer of Massey's has been off quite a bit so far. Had UConn over Iowa by 11 or 12 for example.

The algorithm becomes more accurate as teams play more games. Conclusions derived with a minimum of data is bound to be inaccurate

 

[YKCornelius]

That computer of Massey's has been off quite a bit so far. Had UConn over Iowa by 11 or 12 for example.

Rocky, depends on how you look at things. Massey is pretty accurate predicting the opposing team's score. To be more specific, in the 12 games so far, Massey has been +/- six points nine times (and +/- one point four times). That's a pretty impressive track record, in my opinion.

(On a side note, in all three of the "outliers": Utah (20 points), DePaul (13 points) , and USC (12 points), Massey predicted the opposition would score more than they actually did.)

Trying to nail down UConn's point production is where things get sporty. Massey, like most of the Boneyard's prognostications, is all over the place with regards to the Huskies' scoring and, hence, so is the resultant MOV.

 

[RockyMTblue2]

The algorithm becomes more accurate as teams play more games. Conclusions derived with a minimum of data is bound to be inaccurate

I know and early in the fall I said I don't bother with Massey until mid-December, but low and behold Iowa. So maybe January 15? Betting spreads tend to be more reliiable. Maybe the site should have a warning label.

 

[RockyMTblue2]

Rocky, depends on how you look at things. Massey is pretty accurate predicting the opposing team's score. To be more specific, in the 12 games so far, Massey has been +/- six points nine times (and +/- one point four times). That's a pretty impressive track record, in my opinion.

(On a side note, in all three of the "outliers": Utah (20 points), DePaul (13 points) , and USC (12 points), Massey predicted the opposition would score more than they actually did.)

Trying to nail down UConn's point production is where things get sporty. Massey, like most of the Boneyard's prognostications, is all over the place with regards to the Huskies' scoring and, hence, so is the resultant MOV.

Do not confuse me with facts. Can't you see I'm on a roll here! No, seriously, there is no explanation for UConn's "sporty" predictions as #1.

 

[Suebird10fan]

I know and early in the fall I said I don't bother with Massey until mid-December, but lo and behold Iowa. So maybe January 15? Betting spreads tend to be more reliiable. Maybe the site should have a warning label.

I find Bart Torvik to be the best at this point, then Sonny Moore. They all heavily discount blowouts (torvik much less so) which is very problematic for Wcbb top teams.

 

[visitingcock]

I know and early in the fall I said I don't bother with Massey until mid-December, but low and behold Iowa. So maybe January 15? Betting spreads tend to be more reliiable. Maybe the site should have a warning label.

That works with many like the Big 10 and Big 12 that have already started conference play. But we will need to look at SEC teams around the first week of February.

 

[cancontent]

I know and early in the fall I said I don't bother with Massey until mid-December, but low and behold Iowa. So maybe January 15? Betting spreads tend to be more reliiable. Maybe the site should have a warning label.

What doesn't come with a warning label nowadays? 😉

 

[NycUcWbbFan]

Rocky, depends on how you look at things. Massey is pretty accurate predicting the opposing team's score. To be more specific, in the 12 games so far, Massey has been +/- six points nine times (and +/- one point four times). That's a pretty impressive track record, in my opinion.

(On a side note, in all three of the "outliers": Utah (20 points), DePaul (13 points) , and USC (12 points), Massey predicted the opposition would score more than they actually did.)

Trying to nail down UConn's point production is where things get sporty. Massey, like most of the Boneyard's prognostications, is all over the place with regards to the Huskies' scoring and, hence, so is the resultant MOV.

Massey’s maximum likelihood estimation has an underlying assumption: that margin of victory (including transformations on MOV, and linear combinations thereof) are normally distributed (equivalently, their error terms are normally distributed).

Central limit theorem normality (i.e. when there are enough representative games, which under the Quad System normally means in the teeth of conference play) also works.

That is why early season Massey is wacky vs late season. And Massey measures season-to-date normality — it doesn’t weight recent games more (unlike, for example Torvik). &1

Last year’s UConn team (with its pronounced pre- and post- South Carolina metamorphosis) would be undervalued by Massey (and NET) vs. Torvik.

In the context of defense and offense to-date by this year’s team, Geno is on record as saying that defense is ahead of the (half-court offense) on a team that is preliminarily comparable to the better UConn teams.

&1 In addition, Torvik (unlike Massey) makes granular adjustments to the weight of blowout games (both make adjustments for blowout scores).

 

[The CDR-Ret]

The other thing that stood out to me is Shea having Vandy at #14.

Both men and women's teams at Vanderbilt and Iowa State are still undefeated, yet three of the four are in double digit rankings. With UConn, I like the #1 and #4 rankings as they are both still improving faster than other teams.

Go Huskies^2!!!

 

[RockyMTblue2]

Massey’s maximum likelihood estimation has an underlying assumption: that margin of victory (including transformations on MOV, and linear combinations thereof) are normally distributed (equivalently, their error terms are normally distributed).

Central limit theorem normality (i.e. when there are enough representative games, which under the Quad System normally means in the teeth of conference play) also works.

That is why early season Massey is wacky vs late season. And Massey measures season-to-date normality — it doesn’t weight recent games more (unlike, for example Torvik). &1

Last year’s UConn team (with its pronounced pre- and post- South Carolina metamorphosis) would be undervalued by Massey (and NET) vs. Torvik.

In the context of defense and offense to-date by this year’s team, Geno is on record as saying that defense is ahead of the (half-court offense) on a team that is preliminarily comparable to the better UConn teams.

&1 In addition, Torvik (unlike Massey) makes granular adjustments to the weight of blowout games (both make adjustments for blowout scores).

!! How do you sleep at night?? 😁 And what the heck is a granular adjustment? No, don't tell me. I need what's left of my innocence.