Doggydaddy's Den: Doggy's Doghouse 11/30/2021

****************************************************

GO UCONN!!!!!

**************************************************

Happy third night of Chanukah!!

**************************************************

Latest commit KK Arnold Highlights!!

**************************************************

Great article in the UConn links section from the inside of this cover.

**************************************************

Sunday is our Olympic Heritage Game! We hear there'll be some legends in the building 👀

Fewer than 200 tickets remain ⬇️ https://t.co/ObzBkQ3hHG pic.twitter.com/8wfkdf8NE3
— UConn Women’s Basketball (@UConnWBB) November 30, 2021

**************************************************

UPSETS IN RED

MOST SCORES HAVE CLICKABLE LINKS TO A GAME STORY

No. 5 Baylor 37 Morehead State 17 at the half

********************************************************

By Phil

Vol fans are right

In DoggyDaddy's recent blog post he said, "I just love how they believe that Tennessee gets treated differently than any other team."

Actually, Vol fans are correct. Let's look at the evidence.

First, a little background.

There are fundamentally two types of ratings. One type is a pure mathematical algorithm, such as RPI, NET or Massey ratings. The other type is human polls, such as the AP and coaches poll. A lot can be said about the differences, but I'll emphasize only three points:

1. Mathematical algorithms are typically terrible with too little data, so the early-season ratings can often sometimes absurd. They tend to get better over the course of the season and are reasonably good at the end of the season.
2. Human based polls have the advantage of incorporating a lot of information, such as injuries and recovery from injuries, that would be very difficult to quantify with an algorithm. They typically outperform pure mathematical algorithms early in the season.
3. The advantages of human polls are often offset by psychology factors, such as crowd behavior, sometimes called contagion theory. Another psychological flaw is the penchant for humans to overweight reputation. (This will be the main point of this post)

I hope DD won't yell at me too much for pointing out the Tennessee used to be really, really good, especially in years which were mostly before players on the current teams were born. This reputation still manifests itself in the rankings.

But don't just take my word for it, let's look at some evidence. Algorithms don't give a hoot about reputation, so let's look at the end of the year Massey rankings and see how they stack up against Tennessee's ratings in recent years.

In 2020–21, Tennessee ended the year ranked number 13 in the AP poll and number 16 in the Coaches Poll. Massey had them at number 20, so seven and four points respectively lower. If this was all I had, it would not mean very much as any single data point can be affected by a number of issues, and it would be statistically inappropriate to draw too much inference from one observation.

In 2019–20, Massey ranked Tennessee at number 30. The human polls had Tennessee ranked as high as number 17 during the season but typically in the 20s. It wasn't until February when the human voters finally threw in the towel and drop them from the ratings. This is a clear example showing how the humans gave more weight to the reputation of Tennessee than the computer algorithm that simply looked at the performance.

In 2018–19, Massey ranked Tennessee at number 34. The human polls had Tennessee ranked as highly as eighth, before finally dropping them from the rankings in late January. Another clear example that humans were giving too much weight to reputation from years past.
When we get back to 2017–18, we are getting to a year in which Tennessee was deservedly a top 25 team. They ended the year ranked number 15 by Massey but look at the human polls. The polls had them as high as seventh during the season, although results or lack thereof caught up to them and the posters had Tennessee at 12th and 17th for the AP and coaches respectively. The average of the final human polls is remarkably close to Massey but note that the human effect of over-weighting reputation had Tennessee well above their final rank most of the season.

In 2017–18, Massey had Tennessee at number 21 at the end of the season, one of the few examples where the Massey algorithm rated Tennessee more highly than the posters at the end of the season. However, note that the posters had Tennessee as high as 13 early in the season, again overweighting reputation, and it was only when the losses mounted that the posters finally gave up the ghost.

I haven't worked at other years, but I won't be surprised if a similar pattern occurs. If we go far enough back, will get into years where Tennessee was pretty good, so I would guess there would be a closer match between the pollsters results and Massey results.

**************************************************

By David in Naples

The latest AP poll ranking WBB teams has UConn tied for 2nd place with North Carolina State (6-1), just behind #1 South Carolina. There are no fewer than 10 undefeated teams in the top 25 list. South Carolina is 7-0, #7 Arizona is also 7-0. The remaining undefeated teams include Iowa, Tennessee, Michigan, Iowa State, Texas A&M and Florida Gulf Coast. Many of these teams will play each other during the regular season, guaranteeing someone a loss. For example, the SEC has 4 of the 10 teams without a loss. They all play each other in their conference schedule.

The same is true for the PAC 12, which has 4 teams in the top 25, although only 1, Arizona (7-0), is undefeated. The Big Ten conference has five teams in the top 25 and three remain undefeated. They are #9 Iowa, #12 Michigan and #18 Ohio State. Those teams with "cupcake" opening schedules will soon face tougher games in conference.

Thursday should start the thinning process as #2 NC State faces #6 Indiana and undefeated #9 Iowa plays against undefeated, but unranked, Duke (6-0). Thursday also has #12 Michigan (7-0) facing #10 Louisville (5-1). Let the losses begin...