Gauging Conference Performance Relative to Seed

Discussion:

(too old to reply)

Con Reeder, unhyphenated American

2013-03-23 04:39:30 UTC

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big 10 11.69 (11-12) 6*: Indiana (1, 3.38, 1*), Minnesota (11, 0.54, 1*),
Michigan (4, 1.49, 1*), Ohio St. (2, 2.42, 1*), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 1*), Illinois (7, 0.84, 1*)

Atlantic 10 4.64 (4-5) 5*: La Salle (13, 0.26, 1*), Temple (9, 0.56, 1*),
Virginia Commonwealth (5, 1.16, 1*), St. Louis (4, 1.49, 1*), Butler (6,
1.17, 1*)

ACC 6.21 (6-7) 3*: Miami (FL) (2, 2.42, 1*), Duke (2, 2.42, 1*), North
Carolina (8, 0.69, 1*), North Carolina St. (8, 0.69, 0)

Pac-12 4.03 (4) 3*: Oregon (12, 0.52, 1*), California (12, 0.52, 1*),
Colorado (10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 1*)

Big East 11.88 (11-12) 3*: Marquette (3, 1.86, 1*), Syracuse (4, 1.49, 1*),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 1*), Georgetown (2, 2.42, 0)

MVC 1.40 (1-2) 2*: Wichita St. (9, 0.56, 1*), Creighton (7, 0.84, 1*)

SEC 2.94 (2-3) 2*: Mississippi (12, 0.52, 1*), Missouri (9, 0.56, 0),
Florida (3, 1.86, 1*)

Mountain West 4.54 (4-5) 2*: San Diego St. (7, 0.84, 1*), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1*), Nevada Las Vegas (5, 1.16, 0)

Big 12 7.33 (7-8) 2*: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1*),
Kansas (1, 3.38, 1*), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

C-USA 1.17 (1-2) 1*: Memphis (6, 1.17, 1*)

Ivy 0.16 (0-1) 1*: Harvard (14, 0.16, 1*)

Atlantic Sun 0.05 (0) 1*: Florida Gulf Coast (15, 0.05, 1*)

West Coast 3.92 (3-4) 1*: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1*)

Conference Left/Bids W L PCT Expected
Big 10 6/7 6 1 0.857 11.69
Atlantic 10 5/5 5 0 1.000 4.64
ACC 3/4 3 1 0.750 6.21
Pac-12 3/5 3 2 0.600 4.03
Big East 3/8 3 5 0.375 11.88
MVC 2/2 2 0 1.000 1.40
SEC 2/3 2 1 0.667 2.94
Mountain West 2/4 2 2 0.500 4.54
Big 12 2/5 2 3 0.400 7.33
C-USA 1/1 1 0 1.000 1.17
Ivy 1/1 1 0 1.000 0.16
Atlantic Sun 1/1 1 0 1.000 0.05
West Coast 1/2 1 1 0.500 3.92
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 32/64 32 32 63.00

Cornelius Cornelius

2013-03-23 08:03:18 UTC

Permalink

On Mar 23, 4:39 am, "Con Reeder, unhyphenated American"

Post by Con Reeder, unhyphenated American
($1 to George Harris)
Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.
I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000
Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).
So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.
Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.
So, without further ado, here are the expected wins for each
Big 10 11.69 (11-12) 6*: Indiana (1, 3.38, 1*), Minnesota (11, 0.54, 1*),
Michigan (4, 1.49, 1*), Ohio St. (2, 2.42, 1*), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 1*), Illinois (7, 0.84, 1*)
Atlantic 10 4.64 (4-5) 5*: La Salle (13, 0.26, 1*), Temple (9, 0.56, 1*),
Virginia Commonwealth (5, 1.16, 1*), St. Louis (4, 1.49, 1*), Butler (6,
1.17, 1*)
ACC 6.21 (6-7) 3*: Miami (FL) (2, 2.42, 1*), Duke (2, 2.42, 1*), North
Carolina (8, 0.69, 1*), North Carolina St. (8, 0.69, 0)
Pac-12 4.03 (4) 3*: Oregon (12, 0.52, 1*), California (12, 0.52, 1*),
Colorado (10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 1*)
Big East 11.88 (11-12) 3*: Marquette (3, 1.86, 1*), Syracuse (4, 1.49, 1*),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 1*), Georgetown (2, 2.42, 0)
MVC 1.40 (1-2) 2*: Wichita St. (9, 0.56, 1*), Creighton (7, 0.84, 1*)
SEC 2.94 (2-3) 2*: Mississippi (12, 0.52, 1*), Missouri (9, 0.56, 0),
Florida (3, 1.86, 1*)
Mountain West 4.54 (4-5) 2*: San Diego St. (7, 0.84, 1*), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1*), Nevada Las Vegas (5, 1.16, 0)
Big 12 7.33 (7-8) 2*: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1*),
Kansas (1, 3.38, 1*), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)
C-USA 1.17 (1-2) 1*: Memphis (6, 1.17, 1*)
Ivy 0.16 (0-1) 1*: Harvard (14, 0.16, 1*)
Atlantic Sun 0.05 (0) 1*: Florida Gulf Coast (15, 0.05, 1*)
West Coast 3.92 (3-4) 1*: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1*)
Conference Left/Bids W L PCT Expected
Big 10 6/7 6 1 0.857 11.69
Atlantic 10 5/5 5 0 1.000 4.64
ACC 3/4 3 1 0.750 6.21
Pac-12 3/5 3 2 0.600 4.03
Big East 3/8 3 5 0.375 11.88
MVC 2/2 2 0 1.000 1.40
SEC 2/3 2 1 0.667 2.94
Mountain West 2/4 2 2 0.500 4.54
Big 12 2/5 2 3 0.400 7.33
C-USA 1/1 1 0 1.000 1.17
Ivy 1/1 1 0 1.000 0.16
Atlantic Sun 1/1 1 0 1.000 0.05
West Coast 1/2 1 1 0.500 3.92
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 32/64 32 32 63.00

So unless Harvard make it to the round of 8, the Atlantic Sun wins.

Con Reeder, unhyphenated American

2013-03-25 04:05:08 UTC

Permalink

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big 10 11.69 (11-12) 10*: Indiana (1, 3.38, 2*), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 2*), Ohio St. (2, 2.42, 2*), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2*), Illinois (7, 0.84, 1)

Big East 11.88 (11-12) 6*: Marquette (3, 1.86, 2*), Syracuse (4, 1.49, 2*),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 2*), Georgetown (2, 2.42, 0)

ACC 6.21 (6-7) 5*: Miami (FL) (2, 2.42, 2*), Duke (2, 2.42, 2*), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)

Pac-12 4.03 (4) 5*: Oregon (12, 0.52, 2*), California (12, 0.52, 1),
Colorado (10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2*)

Atlantic Sun 0.05 (0) 2*: Florida Gulf Coast (15, 0.05, 2*)

MVC 1.40 (1-2) 3*: Wichita St. (9, 0.56, 2*), Creighton (7, 0.84, 1)

SEC 2.94 (2-3) 3*: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0),
Florida (3, 1.86, 2*)

Atlantic 10 4.64 (4-5) 6*: La Salle (13, 0.26, 2*), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)

Big 12 7.33 (7-8) 3*: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1),
Kansas (1, 3.38, 2*), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)

Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)

West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)

Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)

Conference Left/Bids W L PCT Expected
Big 10 4/7 10 3 0.769 11.69
Big East 3/8 6 5 0.545 11.88
ACC 2/4 5 2 0.714 6.21
Pac-12 2/5 5 3 0.625 4.03
Atlantic Sun 1/1 2 0 1.000 0.05
MVC 1/2 3 1 0.750 1.40
SEC 1/3 3 2 0.600 2.94
Atlantic 10 1/5 6 4 0.600 4.64
Big 12 1/5 3 4 0.429 7.33
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 16/64 48 48 63.00

--
Why is it more moral for a federal bureaucrat in a state-supplied SUV to
shut down an offshore oil rig on grounds that it is too dangerous for
the environment than for a private individual to risk his own capital to
find some sort of new fuel to power his government's SUV fleet?
-- Victor Davis Hanson

Con Reeder, unhyphenated American

2013-03-30 04:57:17 UTC

Permalink

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big East 11.88 (11-12) 9*: Marquette (3, 1.86, 3*), Syracuse (4, 1.49, 3*),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 3*), Georgetown (2, 2.42, 0)

Big 10 11.69 (11-12) 12*: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 3*), Ohio St. (2, 2.42, 3*), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)

MVC 1.40 (1-2) 4*: Wichita St. (9, 0.56, 3*), Creighton (7, 0.84, 1)

SEC 2.94 (2-3) 4*: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0),
Florida (3, 1.86, 3*)

ACC 6.21 (6-7) 6*: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3*), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)

Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)

Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)

C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)

Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)

Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)

Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)

Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)

Conference Left/Bids W L PCT Expected
Big East 3/8 9 5 0.643 11.88
Big 10 2/7 12 5 0.706 11.69
MVC 1/2 4 1 0.800 1.40
SEC 1/3 4 2 0.667 2.94
ACC 1/4 6 3 0.667 6.21
Atlantic Sun 0/1 2 1 0.667 0.05
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 8/64 56 56 63.00

--
There comes a time when you should stop expecting other people to make
a big deal about your birthday. That time is age 12. -- Dave Barry

Con Reeder, unhyphenated American

2013-04-01 04:15:24 UTC

Permalink

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big East 11.88 (11-12) 11*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4*),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 4*), Georgetown (2, 2.42, 0)

MVC 1.40 (1-2) 5*: Wichita St. (9, 0.56, 4*), Creighton (7, 0.84, 1)

Big 10 11.69 (11-12) 13*: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 4*), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)

Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)

ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)

SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)

Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)

C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)

Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)

Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)

Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)

Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)

Conference Left/Bids W L PCT Expected
Big East 2/8 11 6 0.647 11.88
MVC 1/2 5 1 0.833 1.40
Big 10 1/7 13 6 0.684 11.69
Atlantic Sun 0/1 2 1 0.667 0.05
ACC 0/4 6 4 0.600 6.21
SEC 0/3 4 3 0.571 2.94
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 4/64 60 60 63.00

--
How far can you open your mind before your brains fall out?

Con Reeder, unhyphenated American

2013-04-07 03:45:53 UTC

Permalink

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big 10 11.69 (11-12) 14*: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 5*), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)

Big East 11.88 (11-12) 12*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 5*), Georgetown (2, 2.42, 0)

MVC 1.40 (1-2) 5: Wichita St. (9, 0.56, 4), Creighton (7, 0.84, 1)

Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)

ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)

SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)

Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)

C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)

Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)

Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)

Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)

Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)

Conference Left/Bids W L PCT Expected
Big 10 1/7 14 6 0.700 11.69
Big East 1/8 12 7 0.632 11.88
MVC 0/2 5 2 0.714 1.40
Atlantic Sun 0/1 2 1 0.667 0.05
ACC 0/4 6 4 0.600 6.21
SEC 0/3 4 3 0.571 2.94
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 2/64 62 62 63.00

--
Find the grain of truth in criticism, chew it, and swallow
it. -- anonymous

Con Reeder, unhyphenated American

2013-04-09 20:12:58 UTC

Permalink

($1 to George Harris)

Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.

I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
thousandth, expected wins for each of the sixteen seeds:

Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000

Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).

So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.

Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.

So, without further ado, here are the expected wins for each
conference in the coming NCAA tournament:

Big East 11.88 (11-12) 13*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 6*), Georgetown (2, 2.42, 0)

MVC 1.40 (1-2) 5: Wichita St. (9, 0.56, 4), Creighton (7, 0.84, 1)

Big 10 11.69 (11-12) 14: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 5), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)

Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)

ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)

SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)

Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)

C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)

Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)

Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)

Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)

West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)

Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)

Conference Left/Bids W L PCT Expected
Big East 1/8 13 7 0.650 11.88
MVC 0/2 5 2 0.714 1.40
Big 10 0/7 14 7 0.667 11.69
Atlantic Sun 0/1 2 1 0.667 0.05
ACC 0/4 6 4 0.600 6.21
SEC 0/3 4 3 0.571 2.94
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 1/64 63 63 63.00

--
The sun, with all those planets revolving around it and dependent on it,
can still ripen a bunch of grapes as if it had nothing else in the
universe to do. -- Galileo

Jim Brown

2013-04-09 20:15:45 UTC

Permalink

On Apr 9, 3:12 pm, "Con Reeder, unhyphenated American"

Post by Con Reeder, unhyphenated American
($1 to George Harris)
Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.
I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000
Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).
So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.
Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.
So, without further ado, here are the expected wins for each
Big East 11.88 (11-12) 13*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 6*), Georgetown (2, 2.42, 0)
MVC 1.40 (1-2) 5: Wichita St. (9, 0.56, 4), Creighton (7, 0.84, 1)
Big 10 11.69 (11-12) 14: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 5), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)
Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)
ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)
SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)
Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)
C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)
Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)
Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)
Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)
West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)
Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)
Conference Left/Bids W L PCT Expected
Big East 1/8 13 7 0.650 11.88
MVC 0/2 5 2 0.714 1.40
Big 10 0/7 14 7 0.667 11.69
Atlantic Sun 0/1 2 1 0.667 0.05
ACC 0/4 6 4 0.600 6.21
SEC 0/3 4 3 0.571 2.94
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 1/64 63 63 63.00
--
The sun, with all those planets revolving around it and dependent on it,
can still ripen a bunch of grapes as if it had nothing else in the
universe to do. -- Galileo

Thanks for the updates Con.

xyzzy

2013-04-09 20:54:54 UTC

Permalink

On Apr 9, 4:12 pm, "Con Reeder, unhyphenated American"

Post by Con Reeder, unhyphenated American
($1 to George Harris)
Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.
I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
Expected wins by seed
-----------------------
1: 3.375 2: 2.420
3: 1.857 4: 1.491
5: 1.161 6: 1.170
7: 0.839 8: 0.688
9: 0.562 10: 0.652
11: 0.545 12: 0.518
13: 0.259 14: 0.161
15: 0.054 16: 0.000
Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).
So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.
Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.
So, without further ado, here are the expected wins for each
Big East 11.88 (11-12) 13*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 6*), Georgetown (2, 2.42, 0)
MVC 1.40 (1-2) 5: Wichita St. (9, 0.56, 4), Creighton (7, 0.84, 1)
Big 10 11.69 (11-12) 14: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 5), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)
Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)
ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)
SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)
Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)
C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)
Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)
Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)
Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)
West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)
Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)
Conference Left/Bids W L PCT Expected
Big East 1/8 13 7 0.650 11.88
MVC 0/2 5 2 0.714 1.40
Big 10 0/7 14 7 0.667 11.69
Atlantic Sun 0/1 2 1 0.667 0.05
ACC 0/4 6 4 0.600 6.21
SEC 0/3 4 3 0.571 2.94
Atlantic 10 0/5 6 5 0.545 4.64
C-USA 0/1 1 1 0.500 1.17
Ivy 0/1 1 1 0.500 0.16
Pac-12 0/5 5 5 0.500 4.03
Big 12 0/5 3 5 0.375 7.33
West Coast 0/2 1 2 0.333 3.92
Mountain West 0/4 2 4 0.333 4.54
Sun Belt 0/1 0 1 0.000 0.00
Horizon 0/1 0 1 0.000 0.16
Summit 0/1 0 1 0.000 0.26
Southern 0/1 0 1 0.000 0.16
Mid-Eastern 0/1 0 1 0.000 0.00
America East 0/1 0 1 0.000 0.05
Big Sky 0/1 0 1 0.000 0.26
Big West 0/1 0 1 0.000 0.05
WAC 0/1 0 1 0.000 0.26
Colonial 0/1 0 1 0.000 0.00
Southland 0/1 0 1 0.000 0.16
Ohio Valley 0/1 0 1 0.000 0.54
Patriot 0/1 0 1 0.000 0.54
MAC 0/1 0 1 0.000 0.52
Metro Atlantic 0/1 0 1 0.000 0.05
Southwestern 0/1 0 1 0.000 0.00
TOTALS 1/64 63 63 63.00
--
The sun, with all those planets revolving around it and dependent on it,
can still ripen a bunch of grapes as if it had nothing else in the
universe to do. -- Galileo

I guess I don't understand these rankings. SEC won 4 and was expected
to win 2.94. ACC won 6 and was expected to win 6.21. Shouldn't the
SEC be ranked above the ACC given those facts, since SEC overachieved
and ACC performed pretty much as expected? Or you just ranked them by
win pct and gave the expected numbers as "oh by the way" information?

Con Reeder, unhyphenated American

2013-04-09 21:13:07 UTC

Permalink

On Apr 9, 4:12?pm, "Con Reeder, unhyphenated American"

Post by Con Reeder, unhyphenated American
($1 to George Harris)
Something that frequently comes up in basketball discussions is the
question of how well a team or a conference performed in the NCAA
tournament, as related to how they are expected to perform. One of the
main stumbling blocks in this pursuit (not that it's ever caused
anyone to refrain) is the lack of any clear measure of expectation for
a team. One of the most frequently proposed measures is a rather
simplistic interpretation based on a team's seeding: a team with a 1
seed is expected to win 4 games, a 2 seed is expected to win 3 games,
a 3 or 4 seed is expected to win 2 games, and so on down the line.
This measure has a number of drawbacks, including the fact that it
accounts for only 60 of the 63 games played in the tournament, but
foremost among them is that it is very unrealistic: it proposes, for
example, that we should expect all 2 seeds to make it to the regional
finals and then lose, and this is something that almost never happens.
I suggest that it is far more reasonable, if we are to judge a team's
relative performance on its seeding, that we should rather compare it
to how other teams with that seeding have performed in the past. Thus,
I have examined the data since the NCAA tournament expanded to 64
teams in 1985, and here are the total and, rounded to the nearest
?Expected wins by seed
-----------------------
?1: 3.375 ? 2: 2.420
?3: 1.857 ? 4: 1.491
?5: 1.161 ? 6: 1.170
?7: 0.839 ? 8: 0.688
?9: 0.562 ?10: 0.652
11: 0.545 ?12: 0.518
13: 0.259 ?14: 0.161
15: 0.054 ?16: 0.000
Of note is that 10 seeds have been somewhat more successful than 9
seeds; avoiding the 1 seed in the second round is important (only
four 9 seeds have ever made the Sweet Sixteen, and one the Elite
Eight, while the numbers for 10 seeds are twenty and seven,
respectively).
So now we can look at the seeds each conference received, and estimate
how many games members of that conference should be expected to win.
Obviously we can't expect a team to win 3.32 games, but we can expect
a team to win 3 or 4 games. Moreover, we can use our good friend
addition to gauge a conference's performance based on the seedings of
its teams. For example, the SEC in 2001 had teams receive the
following seeds: 2, 3, 3, 7, 8, 8. Using this method, we would expect
the conference to amass approximately
2.42+1.85+1.85+0.85+0.66+0.66 = 8.29, or, say, 8-9 wins.
So, without further ado, here are the expected wins for each
Big East 11.88 (11-12) 13*: Marquette (3, 1.86, 3), Syracuse (4, 1.49, 4),
Villanova (9, 0.56, 0), Cincinnati (10, 0.65, 0), Notre Dame (7, 0.84, 0),
Pittsburgh (8, 0.69, 0), Louisville (1, 3.38, 6*), Georgetown (2, 2.42, 0)
MVC 1.40 (1-2) 5: Wichita St. (9, 0.56, 4), Creighton (7, 0.84, 1)
Big 10 11.69 (11-12) 14: Indiana (1, 3.38, 2), Minnesota (11, 0.54, 1),
Michigan (4, 1.49, 5), Ohio St. (2, 2.42, 3), Wisconsin (5, 1.16, 0),
Michigan St. (3, 1.86, 2), Illinois (7, 0.84, 1)
Atlantic Sun 0.05 (0) 2: Florida Gulf Coast (15, 0.05, 2)
ACC 6.21 (6-7) 6: Miami (FL) (2, 2.42, 2), Duke (2, 2.42, 3), North
Carolina (8, 0.69, 1), North Carolina St. (8, 0.69, 0)
SEC 2.94 (2-3) 4: Mississippi (12, 0.52, 1), Missouri (9, 0.56, 0), Florida
(3, 1.86, 3)
Atlantic 10 4.64 (4-5) 6: La Salle (13, 0.26, 2), Temple (9, 0.56, 1),
Virginia Commonwealth (5, 1.16, 1), St. Louis (4, 1.49, 1), Butler (6,
1.17, 1)
C-USA 1.17 (1-2) 1: Memphis (6, 1.17, 1)
Ivy 0.16 (0-1) 1: Harvard (14, 0.16, 1)
Pac-12 4.03 (4) 5: Oregon (12, 0.52, 2), California (12, 0.52, 1), Colorado
(10, 0.65, 0), UCLA (6, 1.17, 0), Arizona (6, 1.17, 2)
Big 12 7.33 (7-8) 3: Oklahoma (10, 0.65, 0), Iowa St. (10, 0.65, 1), Kansas
(1, 3.38, 2), Kansas St. (4, 1.49, 0), Oklahoma St. (5, 1.16, 0)
West Coast 3.92 (3-4) 1: St. Marys (11, 0.54, 0), Gonzaga (1, 3.38, 1)
Mountain West 4.54 (4-5) 2: San Diego St. (7, 0.84, 1), New Mexico (3,
1.86, 0), Colorado St. (8, 0.69, 1), Nevada Las Vegas (5, 1.16, 0)
Conference ? ? ? ? ? Left/Bids ?W ? L ? ?PCT ? ?Expected
Big East ? ? ? ? ? ? ? ?1/8 ? ?13 ? 7 ? ?0.650 ?11.88
MVC ? ? ? ? ? ? ? ? ? ? 0/2 ? ? 5 ? 2 ? ?0.714 ? 1.40
Big 10 ? ? ? ? ? ? ? ? ?0/7 ? ?14 ? 7 ? ?0.667 ?11.69
Atlantic Sun ? ? ? ? ? ?0/1 ? ? 2 ? 1 ? ?0.667 ? 0.05
ACC ? ? ? ? ? ? ? ? ? ? 0/4 ? ? 6 ? 4 ? ?0.600 ? 6.21
SEC ? ? ? ? ? ? ? ? ? ? 0/3 ? ? 4 ? 3 ? ?0.571 ? 2.94
Atlantic 10 ? ? ? ? ? ? 0/5 ? ? 6 ? 5 ? ?0.545 ? 4.64
C-USA ? ? ? ? ? ? ? ? ? 0/1 ? ? 1 ? 1 ? ?0.500 ? 1.17
Ivy ? ? ? ? ? ? ? ? ? ? 0/1 ? ? 1 ? 1 ? ?0.500 ? 0.16
Pac-12 ? ? ? ? ? ? ? ? ?0/5 ? ? 5 ? 5 ? ?0.500 ? 4.03
Big 12 ? ? ? ? ? ? ? ? ?0/5 ? ? 3 ? 5 ? ?0.375 ? 7.33
West Coast ? ? ? ? ? ? ?0/2 ? ? 1 ? 2 ? ?0.333 ? 3.92
Mountain West ? ? ? ? ? 0/4 ? ? 2 ? 4 ? ?0.333 ? 4.54
Sun Belt ? ? ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.00
Horizon ? ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.16
Summit ? ? ? ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.26
Southern ? ? ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.16
Mid-Eastern ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.00
America East ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.05
Big Sky ? ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.26
Big West ? ? ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.05
WAC ? ? ? ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.26
Colonial ? ? ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.00
Southland ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.16
Ohio Valley ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.54
Patriot ? ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.54
MAC ? ? ? ? ? ? ? ? ? ? 0/1 ? ? 0 ? 1 ? ?0.000 ? 0.52
Metro Atlantic ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.05
Southwestern ? ? ? ? ? ?0/1 ? ? 0 ? 1 ? ?0.000 ? 0.00
TOTALS ? ? ? ? ? ? ? ? ?1/64 ? 63 ?63 ? ? ? ? ? 63.00
--
The sun, with all those planets revolving around it and dependent on it,
can still ripen a bunch of grapes as if it had nothing else in the
universe to do. -- Galileo

I rank them based on winning percentage, with the exception that
conferences with an undefeated team remaining go to the top. I suppose
I could rank them differently on the final one, with ratio of wins
compared to expected wins being the primary sort factor.

I'll look at making my program more intelligent next year. Assuming
there *is* a next year here... 8-)

--
Experience is what allows you to recognize a mistake the second time you
make it. -- unknown