Bracketmatrix 2021-22

#201
#201
bleedingTNorange said:
Updated seed list with results…

8. Duke L
———
9. Houston
10. Texas Tech
11. Wisconsin W
12. UCLA W
———
13. Illinois L
14. Villanova W
15. Providence
16. Michigan State L
———
17. Tennessee
18. Marquette L
19. Ohio State
20. Texas W
———
21. Alabama
22. Xavier
23. LSU W
24. Connecticut W
———
25. Iowa State L
26. St. Mary’s L
27. USC W
28. Indiana L
Click to expand...
Kansas also lost on Monday, any chance they would slip to a 3 seed with Duke?
 
#202
#202
bleedingTNorange said:
2021-22 Bracketology, Vol. 1: where Tennessee stands, how they could rise/fall, and best/worst NCAA Tournament draws

This is an absolute hell of a read from Will Warren for those of you who follow him, really breaks down a bunch of things, few highlights….

-96% chance we are 1-6 seed.
-60% chance of 3-5 seed.
-most probable to be 4 seed.

19-11(10-8): 6.9 seed
20-10(11-7): 5.5 seed
21-9(12-6): 4.6 seed
22-8(13-5): 3.8 seed
23-7(14-4): 2.8 seed
Click to expand...
T-rankology is among the worst in terms of accuracy on the matrix. Incc is even worse. And TeamRankings is downright dreadful as basically all statistical-based methodologies are. I wouldn’t put much stock in any of these
 
#204
#204
golfballs said:
T-rankology is among the worst in terms of accuracy on the matrix. Incc is even worse. And TeamRankings is downright dreadful as basically all statistical-based methodologies are. I wouldn’t put much stock in any of these
Click to expand...
They seem pretty accurate to what I would’ve blindly suggested…where do you have those projections as being?
 
#206
#206
Also, his “goat draw” won’t happen. One of the bracketing principles is to not put 2 teams from the same conference in the first 4 lines of a region.
 
#208
#208
bleedingTNorange said:
That wasn’t what I was asking, I was saying what would you say each of those numbers should be since you disagree with his take?
Click to expand...
I disagree with his assumption that those are accurate projections and therefore the level of precision is so low that it makes it basically a pointless exercise. Would be better off using historical track record of the matrix this far out from selection Sunday and giving a confidence interval to each position on the s curve.
 
#209
#209
golfballs said:
I disagree with his assumption that those are accurate projections and therefore the level of precision is so low that it makes it basically a pointless exercise. Would be better off using historical track record of the matrix this far out from selection Sunday and giving a confidence interval to each position on the s curve.
Click to expand...
I think they’re as accurate as anything can be this far out…what seed do you think is more accurate than those listed for each outcome? Post your seed for each record and we can compare come selection Sunday.
 
#210
#210
bleedingTNorange said:
I think they’re as accurate as anything can be this far out…what seed do you think is more accurate than those listed for each outcome? Post your seed for each record and we can compare come selection Sunday.
Click to expand...
So….. not accurate.

And why would I do that when I said the level of precision is so low that it’s basically pointless. He’s assigning seeds to the precision of a tenth of a point! Lol. That only makes sense if you also assign a confidence interval along with it. Like a 19-11 record projects as a 6.9 seed +/- 2.5.
 
#211
#211
That level of precision can be used as a relative ranking. But at that point just use the matrix which is consistently more accurate in total than these individual statistical models
 
#212
#212
golfballs said:
So….. not accurate.

And why would I do that when I said the level of precision is so low that it’s basically pointless. He’s assigning seeds to the precision of a tenth of a point! Lol. That only makes sense if you also assign a confidence interval along with it. Like a 19-11 record projects as a 6.9 seed +/- 2.5.
Click to expand...
The point is projection, same as all bracketology is this far out. He says 22-8(13-5) he calculates as a 3.8 seed, meaning 4 seed as most likely with possibility of a 3 seed, maybe you disagree with that take but I don’t, that seems pretty spot on to me.
 
#213
#213
bleedingTNorange said:
The point is projection, same as all bracketology is this far out. He says 22-8(13-5) he calculates as a 3.8 seed, meaning 4 seed as most likely with possibility of a 3 seed, maybe you disagree with that take but I don’t, that seems pretty spot on to me.
Click to expand...
We’re arguing on two different levels. You’re just using the results of the magical number machine to validate your own preconceptions. I’m arguing that the underlying assumptions in the machine are flawed and the outputs can’t be expected to be accurate. And in fact they have a track record of not being so.
 
  • Like
Reactions: berryvol
#214
#214
golfballs said:
We’re arguing on two different levels. You’re just using the results of the magical number machine to validate your own preconceptions. I’m arguing that the underlying assumptions in the machine are flawed and the outputs can’t be expected to be accurate. And in fact they have a track record of not being so.
Click to expand...
I guess the post will be easy to find and we will be able to determine whether or not it was accurate come selection Sunday 🤷🏻‍♂️
 
#215
#215
bleedingTNorange said:
I guess the post will be easy to find and we will be able to determine whether or not it was accurate come selection Sunday 🤷🏻‍♂️
Click to expand...
I already posted the link above. The seeding probabilities he’s basing this article on already have a track record. It’s not good.
 
#216
#216
This methodology is basically the same that Team Rankings employs. And they rank 130/135 on the bracket matrix. As you should know by now I’m a big stats guy and I use all these sites, but I also know their limitations. And this is one of them. Using stats to predict such a highly variable game like basketball is one thing, but now you’re layering on predicting how highly variable humans will interpret the results of this highly variable game using statistics. Not saying it’s not a fun exercise, but don’t event try to portend a level of precision down to the tenth of a seed. Lmao. At least provide a disclaimer to readers that is basically statistical socery. I think most of the stat nerds who do this want to convince others of their statistical wizardry. Or they know that if they published appropriately vague outcomes nobody would even pay them any attention
 
#217
#217
I take the “down to the tenth” is basically providing the disclaimer that it’s not exact, or a guarantee, no?

This is essentially how I read it…
19-11(10-8): 6-8 seed
20-10(11-7): 5-6 seed
21-9(12-6): 4-5 seed
22-8(13-5): 3-4 seed
23-7(14-4): 2-3 seed

If you feel any of those projections are likely off base, or have a decent probability of being wrong they say that, but to me I think the above looks about right?
 
#218
#218
bleedingTNorange said:
I take the “down to the tenth” is basically providing the disclaimer that it’s not exact, or a guarantee, no?

This is essentially how I read it…
19-11(10-8): 6-8 seed
20-10(11-7): 5-6 seed
21-9(12-6): 4-5 seed
22-8(13-5): 3-4 seed
23-7(14-4): 2-3 seed

If you feel any of those projections are likely off base, or have a decent probability of being wrong they say that, but to me I think the above looks about right?
Click to expand...
Thats better 😁
I’ve spent a lot of breath on something I’ve said is kind of pointless. I work in statistical forecasting and its kind of a petty thing but strikes a small nerve of mine (not bracket projections, but the subject of precision/accuracy)
 
Last edited:
  • Like
Reactions: MinisterofDef#92 and Lankykong
#219
#219
golfballs said:
Thats better 😁
I’ve spent a lot of breath on something I’ve said is kind of pointless. I work in statistical forecasting and its kind of a petty thing but strikes a small nerve of mine (not bracket projections, but the subject of precision/accuracy)
Click to expand...
I didn’t take his story to mean he was guaranteeing a 3.8 seed or anything, I figured common sense given the subject would mean there’s some wiggle room which I laid out above.
 
  • Like
Reactions: Lankykong
#220
#220
bleedingTNorange said:
I didn’t take his story to mean he was guaranteeing a 3.8 seed or anything, I figured common sense given the subject would mean there’s some wiggle room which I laid out above.
Click to expand...
Haha, I hope he’s not guaranteeing it! Because there is no such thing as a 3.8 seed. That’s one reason it’s a little ridiculous. That value suggest there’s a probability distribution but it’s not known to the reader. What is it? It’s pretty critical to understanding the projection! Teamranking gives an overall distribution curve. (Remember their bracketology is one of the least accurate). But at least there’s a sense of the range of outcomes which makes it more useful. Look how wide that is. While a 4 seed is the most probable it still only has a 20% chance of happening according to their model. And consider how wide this range is given they only have 6 potential outcomes for our final regular season record. That means for any given outcome the range of seeds likely has an even fatter-tail distribution. AC8165A0-CC2B-41AD-8EFC-3C4719339738.jpeg
 
Last edited:
#221
#221
8. Duke L
———
9. Houston L
10. Texas Tech L
11. Wisconsin W
12. UCLA W
———
13. Illinois L
14. Villanova W
15. Providence
16. Michigan State L
———
17. Tennessee W
18. Marquette L
19. Ohio State L
20. Texas W
———
21. Alabama W
22. Xavier L
23. LSU W
24. Connecticut W
———
25. Iowa State L
26. St. Mary’s L
27. USC W
28. Indiana L
 
  • Like
Reactions: Willy14 and golfballs
#222
#222
bleedingTNorange said:
8. Duke L
———
9. Houston L
10. Texas Tech L
11. Wisconsin W
12. UCLA W
———
13. Illinois L
14. Villanova W
15. Providence
16. Michigan State L
———
17. Tennessee W
18. Marquette L
19. Ohio State L
20. Texas W
———
21. Alabama W
22. Xavier L
23. LSU W
24. Connecticut W
———
25. Iowa State L
26. St. Mary’s L
27. USC W
28. Indiana L
Click to expand...

Pretty great week for Tennessee so far. Beat Vandy and it’ll be a perfect week.
 
  • Like
Reactions: Willy14 and golfballs
New posts
You must log in or register to reply here.

VN Store



Back
Top