PDA

View Full Version : An Invitation (or Challenge) to Numbers Crunchers



SLOZag
03-14-2014, 03:17 PM
Often I read a post where someone claims Gonzaga (or Coach Few) has over the years "underperformed during March Madness." I realize there may be several ways to measure "underperformance." Here is one way we might measure it: Calculate a weighted winning average during March Madness over a fixed period of years -- 5, 10, 15, etc. So for example, if the time period chosen is 10 years, and a team gets to the big dance 8 times and accumulates 12 wins during that time, it's weighted average would be 12/10 = 1.2 Another team getting to the big dance 4 times and compiling 12 wins would compile an identical average. And to complete this example, a third team getting to the big dance 10 times and accumulating 13 wins would compile a weighted average of 1.3 -- 13/10 = 1.3. However, this approach merely measures consistency of achievement during March Madness, in my view.

The term "underperformance" suggests something else -- achieving less than what a team "should have accomplished during March Madness." Could this be measured by analyzing the original seeding done by the NCAA over a fixed period of years -- 5, 10, 15, etc.? To explain: If the time period chosen is 10 years, and team A has gone to the dance 8 years, then for each of those 8 years you would compare that team's number of wins during March Madness against the average number of wins achieved by the other teams in that same bracket. Example: As a 7 seed, Team A wins 2 games in 2004, whereas the other three 7 seeds that year win 0, 1, and 2 games, for an average of 1 game per team. Hence Team A "overachieved" by 1 game in 2004. Repeat for the other 7 years when Team A was at the big dance, and add together the scores. The result: a composite score that represents Team A's underperformance/overperformance against similar seeds during March Madness over those ten years. You could do the same thing for any team, and compare their relative underperformance/overperformance.

While this measure ignores teams whose underperformance is so great that they don't make it to the big dance, it certainly is one way to measure a team's underperformance/overperformance during March Madness. What do you think? Does anyone have a firm opinion regarding a better way to measure "underperformance/overperformance" during March Madness?

Anyone want to take a crack at seeing where Gonzaga stacks up when compared to other teams regarding "underperformance/overperformance" during March Madness?

wnczagfan
03-14-2014, 04:42 PM
Often I read a post where someone claims Gonzaga (or Coach Few) has over the years "underperformed during March Madness." I realize there may be several ways to measure "underperformance." Here is one way we might measure it: Calculate a weighted winning average during March Madness over a fixed period of years -- 5, 10, 15, etc. So for example, if the time period chosen is 10 years, and a team gets to the big dance 8 times and accumulates 12 wins during that time, it's weighted average would be 12/10 = 1.2 Another team getting to the big dance 4 times and compiling 12 wins would compile an identical average. And to complete this example, a third team getting to the big dance 10 times and accumulating 13 wins would compile a weighted average of 1.3 -- 13/10 = 1.3. However, this approach merely measures consistency of achievement during March Madness, in my view.

The term "underperformance" suggests something else -- achieving less than what a team "should have accomplished during March Madness." Could this be measured by analyzing the original seeding done by the NCAA over a fixed period of years -- 5, 10, 15, etc.? To explain: If the time period chosen is 10 years, and team A has gone to the dance 8 years, then for each of those 8 years you would compare that team's number of wins during March Madness against the average number of wins achieved by the other teams in that same bracket. Example: As a 7 seed, Team A wins 2 games in 2004, whereas the other three 7 seeds that year win 0, 1, and 2 games, for an average of 1 game per team. Hence Team A "overachieved" by 1 game in 2004. Repeat for the other 7 years when Team A was at the big dance, and add together the scores. The result: a composite score that represents Team A's underperformance/overperformance against similar seeds during March Madness over those ten years. You could do the same thing for any team, and compare their relative underperformance/overperformance.

While this measure ignores teams whose underperformance is so great that they don't make it to the big dance, it certainly is one way to measure a team's underperformance/overperformance during March Madness. What do you think? Does anyone have a firm opinion regarding a better way to measure "underperformance/overperformance" during March Madness?

Anyone want to take a crack at seeing where Gonzaga stacks up when compared to other teams regarding "underperformance/overperformance" during March Madness?

Could you simply look at what the seed number was, compared to the other
teams you play. i.e. how many games against teams with a lower seed did you lose (underachieve) versus how many teams with a higher seed did you beat (overachieve)?

LongIslandZagFan
03-14-2014, 04:44 PM
Often I read a post where someone claims Gonzaga (or Coach Few) has over the years "underperformed during March Madness." I realize there may be several ways to measure "underperformance." Here is one way we might measure it: Calculate a weighted winning average during March Madness over a fixed period of years -- 5, 10, 15, etc. So for example, if the time period chosen is 10 years, and a team gets to the big dance 8 times and accumulates 12 wins during that time, it's weighted average would be 12/10 = 1.2 Another team getting to the big dance 4 times and compiling 12 wins would compile an identical average. And to complete this example, a third team getting to the big dance 10 times and accumulating 13 wins would compile a weighted average of 1.3 -- 13/10 = 1.3. However, this approach merely measures consistency of achievement during March Madness, in my view.

The term "underperformance" suggests something else -- achieving less than what a team "should have accomplished during March Madness." Could this be measured by analyzing the original seeding done by the NCAA over a fixed period of years -- 5, 10, 15, etc.? To explain: If the time period chosen is 10 years, and team A has gone to the dance 8 years, then for each of those 8 years you would compare that team's number of wins during March Madness against the average number of wins achieved by the other teams in that same bracket. Example: As a 7 seed, Team A wins 2 games in 2004, whereas the other three 7 seeds that year win 0, 1, and 2 games, for an average of 1 game per team. Hence Team A "overachieved" by 1 game in 2004. Repeat for the other 7 years when Team A was at the big dance, and add together the scores. The result: a composite score that represents Team A's underperformance/overperformance against similar seeds during March Madness over those ten years. You could do the same thing for any team, and compare their relative underperformance/overperformance.

While this measure ignores teams whose underperformance is so great that they don't make it to the big dance, it certainly is one way to measure a team's underperformance/overperformance during March Madness. What do you think? Does anyone have a firm opinion regarding a better way to measure "underperformance/overperformance" during March Madness?

Anyone want to take a crack at seeing where Gonzaga stacks up when compared to other teams regarding "underperformance/overperformance" during March Madness?

This has been done before in another thread at lest relative to GU. Try searching for it.

wnczagfan
03-14-2014, 04:45 PM
By the way, if you look at the brackets I fill out, we underachieve every year we don't win the National Championship. ;)

SLOZag
03-14-2014, 04:54 PM
By the way, if you look at the brackets I fill out, we underachieve every year we don't win the National Championship. ;)

Purrrrrfect reply!

exclusivelee
03-14-2014, 05:59 PM
http://wp.bracketscience.com/?p=760


Revisiting the “Law of the Snake-bit Coach”

All this begs the question: who are the active snake-bit coaches? Here’s a list, ordered first by the number of tourney appearances, then by PASE:


Fran Dunphy, Temple (15 Trips | 10.7 Seed | -0.345 PASE)
Mark Few, Gonzaga (14 Trips | 6.6 Seed | -0.175 PASE)
Mike Brey, Notre Dame (11 Trips | 7.3 Seed | -0.440 PASE)
Dana Altman, Oregon (9 Trips | 9.6 Seed | -0.287 PASE)
Stew Morrill, Utah State (9 Trips | 12.9 Seed | -0.271 PASE)
Kevin Stallings, Vanderbilt (8 Trips | 6.3 Seed | -0.345 PASE)
Tim Floyd, Texas El Paso (8 Trips | 7.6 Seed | 0.012 PASE)
Ben Braun, Rice (8 Trips | 8.8 Seed | 0.164 PASE)
Oliver Purnell, DePaul (7 Trips | 8.2 Seed | -0.819 PASE)
Steve Alford, UCLA (7 Trips | 6.1 Seed | -0.531 PASE)
Leonard Hamilton, Florida State (7 Trips | 6.6 Seed | -0.331 PASE)
Lorenzo Romar, Washington (7 Trips | 6.4 Seed | -0.092 PASE)
Matt Painter, Purdue (7 Trips | 6.6 Seed | 0.073 PASE)
Herb Sendek, Arizona State (7 Trips | 8.1 Seed | 0.107 PASE)
Dave Rose, BYU (6 Trips | 8.0 Seed | -0.320 PASE)
Rick Byrd, Belmont (6 Trips | 13.8 Seed | -0.195 PASE)
Cy Alexander, North Carolina A&T (6 Trips | 15.5 Seed | -0.034 PASE)
Cliff Ellis, Coastal Carolina (6 Trips | 6.0 Seed | -0.010 PASE)
Jim Crews, St. Louis (5 Trips | 9.6 Seed | -0.290 PASE)
Randy Bennett, St. Mary’s (5 Trips | 9.6 Seed | -0.255 PASE)
Todd Bozeman, Morgan State (5 Trips | 10.6 Seed | -0.193 PASE)
Greg McDermott, Creighton (5 Trips | 10.0 Seed | -0.171 PASE)
Mick Cronin, Cincinnati (5 Trips | 9.6 Seed | -0.133 PASE)
Trent Johnson, TCU (5 Trips | 8.0 Seed | 0.117 PASE)
Fran McCaffery, Iowa (5 Trips | 13.4 Seed | 0.176 PASE)
Mark Turgeon, Maryland (5 Trips | 7.4 Seed | 0.209 PASE)


http://wp.bracketscience.com/?p=81


PASE is the primary metric I use to assess the degree to which a coach, team, conference or factor exceeds or falls short of expectations. PASE stands for Performance Against Seed Expectations. Over the 28 years of the 64-team era, each of the seeds has recorded an average number of wins. The 112 top seeds, for instance, have won 378 games since 1985—or 3.375 per team. When a team wins four games as a top seed, therefore, they beat expectations by 0.625 games. Using the averages of each seed, you can tally the positive or negative differences between actual and expected wins for any attribute you want to study—whether it’s an individual coach, teams with more than five straight bids, teams with all-Americans, et cetera. The total of these differences is then divided by the number of appearances to arrive at an average number of games the attribute either over-performed or under-performed per tournament.