Mike Porter
Postdoc
Posts: 3619
Loc: Los Angeles, CA
Reg: 11-21-04
|
Re: Ivy Title/Tourney Odds 02-21-17 09:57 PM - Post#222955
In response to SomeGuy
Completely with you SG. I'm most concerned with Cornell first and think they'll come out fired up at home wanting to make up for the blowout at the Palestra. That will be a tough game.
|
SomeGuy
Professor
Posts: 6418
Reg: 11-22-04
|
Re: Ivy Title/Tourney Odds 02-22-17 09:10 AM - Post#222977
In response to Mike Porter
They'll look a little different, too. Instead of the 4 guard they've run all season, this weekend they went to a more traditional lineup with 2 bigs to start games. So the freshman Warren will play more, which may enable them to matchup a little better when we play with 2 bigs.
They also know that they won the 2nd half at the Palestra pressing and chucking 3s. So we could see a high variance approach (which, unfortunately, can lead to high variance results).
|
mountainred
Masters Student
Posts: 514
Age: 57
Loc: Charleston, WV
Reg: 04-11-10
|
02-22-17 10:22 AM - Post#222981
In response to SomeGuy
I hope Earl doesn't try to draw any grand conclusions from what "worked" in garbage time to what would work while the game is still on the line. After 24 games, it is clear the Big Red isn't a good three point shooting team. When you are #278 in 3 point shooting there will be the occasional hot night, but the bell curve of results is mostly bad.
|
PennFan10
Postdoc
Posts: 3590
Reg: 02-15-15
|
02-24-17 03:36 PM - Post#223257
In response to mrjames
reposting some great tweets by Mr James on tiebreakers:
Mike Jamesâ€@ivybball
Penn is the favorite right now. A win over COL would make it an overwhelming favorite but anything other than a COL sweep is fine or better.
Mike Jamesâ€@ivybball
In the vast, vast majority of scenarios, the 2/3 game is Harvard-Yale & Princeton is the 1, opening with, well, that's where things get fun.
Mike Jamesâ€@ivybball
Of course, there are also some crazier scenarios where a 3-7 team wins out, but let's treat those with the reasonableness they deserve.
Mike Jamesâ€@ivybball
So, for Yale, 8-6 is obviously in. 7-7 would be good so long as the Penn-Columbia loser loses one more game.
Mike Jamesâ€@ivybball
Given that Yale already is 0-2 vs PRIN & would likely be 0-2 vs HAR for this to be an issue, record vs other qualified teams would be rough.
Mike Jamesâ€@ivybball
If Yale needs a tiebreaker to qualify, it will have won at <= 1 game down the stretch. If that game isn't vs COL, it will split H2H breaker.
Mike Jamesâ€@ivybball
Harvard is pretty locked into the 2/3 game unless it loses out (& loses breakers) or makes up 2 gms (prob 3 given tiebreakers) on Princeton.
Mike Jamesâ€@ivybball
Princeton will be the 1-seed at 12-2 or better unless Harvard wins out & the other team the Tigers lose to isn't Columbia & gets the 4-seed.
Mike Jamesâ€@ivybball
With less than 7 hrs to tip, let's reset the Ivy race & provide the cheat sheet for understanding the impacts of various games this weekend.
|
PennFan10
Postdoc
Posts: 3590
Reg: 02-15-15
|
02-24-17 03:38 PM - Post#223258
In response to PennFan10
I wonder how "reasonable" a 3-7 team running the table is/isn't. Can't be any more/less likely than Penn going from 0-6 to 7-7
Cornell and Dartmouth are still in this thing.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
02-24-17 03:54 PM - Post#223261
In response to PennFan10
Here are the "win out" odds for each team:
Brown (6 wins): 1.7%
Columbia (8 wins): 1.9%
Cornell (7 wins): 0.6%
Dartmouth (7 wins): 0.1%
Harvard (12 wins): 5.4%
Penn (8 wins): 15.7%
Princeton (14 wins): 40.8%
Yale (10 wins): 10.4%
So, every team is expected to lose at least once and only Princeton is remotely close to tipping over to going undefeated the rest of the way being more likely than not.
That being said, there is a ~1/3 chance that one of the seven non-Princeton teams will pull off a 4-0 final stretch. And when you throw in Princeton, there's a ~2/3 chance that some team will go 4-0 down the stretch (obviously most likely to be the Tigers).
|
TigerFan
PhD Student
Posts: 1892
Reg: 11-21-04
|
02-24-17 04:03 PM - Post#223265
In response to mrjames
Mike, Wish you had time to discuss variance and the Quakers last night on the vine. Is there a statistical model predicting the Quakers will revert to their norm or is that just a well educated hunch? (is that sufficient bait for you?)
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
02-24-17 04:55 PM - Post#223278
In response to TigerFan
I always manage to destroy that podcast. Feel so bad every single time I go on there.
I'm relying on the work of others here, but essentially, if you were to recursively walk through each team's ratings progression throughout the year, you would find that after any given game, you would be better off predicting that a team moves towards its YTD performance average rather than further away from it.
There are definite exceptions to this. There have been teams that have just progressively increased their ranking all season (Kyle Smith's new squad, for instance). But that's the exception, not the rule. Generally, a short term vacillation in performance tends to regress back toward the dominant YTD trend.
My hypothesis on this is that style-driven improvements or declines are more likely to sustain and not regress than improvements or declines driven by elements more influenced by luck or variance (jumper shooting for and against, free throw defense, non-steal TOs). Why I'm so suspicious of Penn's sudden surge is that its eFG% in the four-game surge has been 48%, 62%, 70% and 59%. It was 57% at La Salle. In the four and three game slides in Jan and Feb it was 49%, 41%, 49% and 49% and 42%, 55% and 47%.
Looking at it from a eFG% for vs. against, you see that during the slides they went:
+11%
-11%
-6%
+3%
-6%
-3%
-16%
And during this run:
+5%
+15%
+19%
+21%
I'm HIGHLY skeptical of runs driven by outsized eFG% differentials. FWIW, Penn's long run eFG% differential is +3% for the year.
|
JadwinGeorge
Senior
Posts: 357
Age: 75
Reg: 12-04-15
|
02-24-17 05:21 PM - Post#223290
In response to mrjames
I wish you could make your Skype connection compatible with Peter Andrews'. Your visits are always interesting and loaded with insight not readily available to casual fans. I enjoyed the discussion "off the air" after we finished the podcast, too. And not just because you predict recognition for Weisz, Cook and Stephens.
|
TheLine
Professor
Posts: 5597
Age: 60
Reg: 07-07-09
|
02-24-17 05:22 PM - Post#223291
In response to mrjames
There has been a personnel change towards players who have been more efficient, like Betley. That has also had a positive effect on the rest of the team - spacing is improved, others are getting less contested shots, Brodeur and Foreman aren't putting up as many low percentage shots.
So I think some of the better efficiency is for real. How much is the question.
|
Streamers
Professor
Posts: 8353
Loc: NW Philadelphia
Reg: 11-21-04
|
02-24-17 05:22 PM - Post#223292
In response to mrjames
I'm HIGHLY skeptical of runs driven by outsized eFG% differentials. FWIW, Penn's long run eFG% differential is +3% for the year.
The data agrees with you; but please consider the fact Penn is effectively putting a different team out on the floor now than they did during 0-6. If there was an anomaly this year, it was the OOC wins against LaSalle and UCF; and I attribute those in large part to inferior coaching/scheme by the opposition.
|
PennFan10
Postdoc
Posts: 3590
Reg: 02-15-15
|
02-24-17 09:09 PM - Post#223370
In response to Streamers
Mike James,
Your twitter stuff is awesome. Thanks for posting it.
|
palestra38
Professor
Posts: 32916
Reg: 11-21-04
|
02-25-17 09:28 AM - Post#223487
In response to mrjames
Here's my question. While Penn may have had a 15.7% chance of winning out (before last night), what was it's chance of winning 5 straight after losing the first 6?
And what is your point spread tonight for both Penn-Columbia and Yale-Dartmouth?
|
SomeGuy
Professor
Posts: 6418
Reg: 11-22-04
|
02-25-17 10:45 AM - Post#223493
In response to palestra38
i'm guessing the odds of losing the 1st six weren't very high either. But the odds of going 5-6 for the first 11 were higher, and overall that's where we are.
|
PennFan10
Postdoc
Posts: 3590
Reg: 02-15-15
|
02-25-17 11:04 AM - Post#223499
In response to SomeGuy
Kenpom has Penn -1 for tonight.
|
Streamers
Professor
Posts: 8353
Loc: NW Philadelphia
Reg: 11-21-04
|
02-25-17 11:07 AM - Post#223501
In response to PennFan10
Vegas has Lions -1 (they know something about Howard)
Note Penn is about 20 slots above Yale in the rankings now.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
Ivy Title/Tourney Odds 03-04-17 01:26 PM - Post#224709
In response to mrjames
Bump?
I'm relying on the work of others here, but essentially, if you were to recursively walk through each team's ratings progression throughout the year, you would find that after any given game, you would be better off predicting that a team moves towards its YTD performance average rather than further away from it.
There are definite exceptions to this. There have been teams that have just progressively increased their ranking all season (Kyle Smith's new squad, for instance). But that's the exception, not the rule. Generally, a short term vacillation in performance tends to regress back toward the dominant YTD trend.
My hypothesis on this is that style-driven improvements or declines are more likely to sustain and not regress than improvements or declines driven by elements more influenced by luck or variance (jumper shooting for and against, free throw defense, non-steal TOs). Why I'm so suspicious of Penn's sudden surge is that its eFG% in the four-game surge has been 48%, 62%, 70% and 59%. It was 57% at La Salle. In the four and three game slides in Jan and Feb it was 49%, 41%, 49% and 49% and 42%, 55% and 47%.
Looking at it from a eFG% for vs. against, you see that during the slides they went:
+11%
-11%
-6%
+3%
-6%
-3%
-16%
And during this run:
+5%
+15%
+19%
+21%
I'm HIGHLY skeptical of runs driven by outsized eFG% differentials. FWIW, Penn's long run eFG% differential is +3% for the year.
|