Prediction

[DC_Area_Panther]

Being retired I had time on my hands so I did the following for amusement-

(1) Listed 2020 pre-season KenPom rankings for all of this year's Pitt opponents.

(2) Assigned numerical ratings to the KenPom rankings based on Sagarin's end of last season numerical ratings for each equivalent Sagarin ranking (Sagarin #5 team = KenPom #5, Sagarin # 121 team = KenPom #121, etc., etc.).

(3) Adjusted the rating numbers based on whether each game was home, away, or neutral. Used Sagarin's 3.13 homecourt advantage from the end of last season to do this.

(4) Then, used the resulting rating numbers to predict a scoring margin for each of this coming season's Pitt games.

The outcome was a prediction that Pitt (hypothetically) would finish 11-0 (OOC) and 6-14 (ACC) for an overall record of 17-14 headed into the ACC tourney.

IMHO, this seems roughly consistent with what some of us have been guessing.

Note: My guess on the OOC outcome has been 9-2; so, 11-0 isn't far off given that this analysis predicts that 2-3 of the 11 wins would be by only be point or two.

To make a comparison for fun, I may do another analysis, but using Sagarin-only pre-season rankings/ratings, once Sagarin releases those ratings (end of the week?).

 

[Joe the Panther Fan]

If you have a 51% chance to win each of 10 games you aren't in any real sense of the word predicted to go 10-0, you are predicted to most likely go 5-5.

 

[DC_Area_Panther]

If you have a 51% chance to win each of 10 games you aren't in any real sense of the word predicted to go 10-0, you are predicted to most likely go 5-5.

I wasn't at all looking at or concerned with probabilities here. Only looking at each game taken individually (in a vacuum so to speak) as to whether we would be favored to win or to lose. Clearly, in the normal course of events we will win some games we would be predicted to lose and lose some we would be predicted to win--but it all should likely even out over the course of the entire season to something like the Ws vs Ls results proposed if the team strength assumptions turn out to be correct. This was just a, for the fun of it, quick and dirty surrogate for the fully accurate probabilistic analysis I am too lazy to perform.

 

[Joe the Panther Fan]

but it all should likely even out over the course of the entire season to something like the Ws vs Ls results proposed if the team strength assumptions turn out to be correct.

No, it won't. That's the point. A team that is a small favorite to win ten games isn't going to have it "even out" and win around ten games. They are most likely going to win five games. If they get a little lucky, six. If they aren't a little lucky, maybe four.

To say that looking at the outcome of your little exercise is that Pitt "would finish 11-0 (OOC)" is dumbing down what Pomeroy and Sagarin are trying to do to the point of meaninglessness.

I don't know about Sagarin, but what Pomeroy is actually saying is that our most likely OOC record, excluding the unknown game against either Bradley or Northwestern, is 8-2. And that 9-1 is slightly more likely than 7-3 as the next two most likely choices. Why isn't it enough to use their ratings the way they were intended, rather than to make up something that they clearly aren't designed to do to try to make things look better?

 

[DT_PITT]

If you have a 51% chance to win each of 10 games you aren't in any real sense of the word predicted to go 10-0, you are predicted to most likely go 5-5.

Perhaps to make the point stronger, if you have an 80% to win each of you 10, you are likely to lose at least one of those games.

 

[DC_Area_Panther]

No, it won't. That's the point. A team that is a small favorite to win ten games isn't going to have it "even out" and win around ten games. They are most likely going to win five games. If they get a little lucky, six. If they aren't a little lucky, maybe four.

To say that looking at the outcome of your little exercise is that Pitt "would finish 11-0 (OOC)" is dumbing down what Pomeroy and Sagarin are trying to do to the point of meaninglessness.

I don't know about Sagarin, but what Pomeroy is actually saying is that our most likely OOC record, excluding the unknown game against either Bradley or Northwestern, is 8-2. And that 9-1 is slightly more likely than 7-3 as the next two most likely choices. Why isn't it enough to use their ratings the way they were intended, rather than to make up something that they clearly aren't designed to do to try to make things look better?

Because I am a Pitt hoops optimist. I always hope that any game we are favored in we will win. We might not win all 11 OOC games but that is a more likely outcome being favored in every game that it would be if we were underdogs in several of those games.

Also, as far as the ACC games piece is concerned we also have some chance of wining (or losing) more ACC games than the number we would currently be considered the favorite in--and, again, the optimist in me hopes that we win more in total than just the ones we are currently favored to win. Finally, at the end of the day, my unscientific gut feeling is that we are going to win 16-17 games (or more) this season with whatever number of games we are favored in but lose being ultimately offset by whatever number of games we are underdogs in that we win.