Thursday, December 25, 2014

College football bowls: Where coaches ranked their opponents

During bowl season, coaches will often say they are excited for the opportunity to travel to a new city and play against a good team in a finishing bowl game. Coaches won't publicly slight the city even if it's not considered a vacation destination and they wouldn't disrespect an opponent either.

However, with public Coaches' poll ballots, we can find out what a coach really thinks about their bowl opponent (though some coaches may hand their ballot to someone else in the athletic department). These ballots were completed just before bowl games and opponents were announced.

Most opposing coaches agreed with each other that one of the two teams in the bowl is superior. Only the Holiday Bowl had the coaches rank the teams in different orders and one of the coaches has been fired (Bo Pelini). As I mentioned in the blog when ballots were released, there is overwhelming evidence for familiarity bias in the polls. Coaches in the same conference rated their teams higher than coaches from other conferences for 33 of 36 teams. There were also quite a few coaches who ranked their own teams extremely high.

Anyway, here's how they voted:

Sugar Bowl:
Urban Meyer (Ohio State): Alabama 1, Ohio State 4
Nick Saban (Alabama): Alabama 1, Ohio State 4

Rose Bowl:
Jimbo Fisher (Florida State): FSU 1, Oregon 3

Cotton Bowl:
Art Briles (Baylor): Baylor 3 (extreme), Michigan State 8
Mark Dantonio (Michigan State): Baylor 5, Michigan State 7

Fiesta Bowl:
Rich Rodriguez (Arizona): Arizona 9, Boise State 17

Orange Bowl
Paul Johnson (Georgia Tech): Mississippi State 8, GT 9

Chick Fil A Peach Bowl:
Gary Patterson (TCU): TCU 3 (extreme), Ole Miss 12

Music City Bowl: Notre Dame won 31-28.
Brian Kelly (Notre Dame): LSU 23, ND NR
Les Miles (LSU): LSU 18 (extreme), ND NR

Citrus Bowl
Jerry Kill (Minnesota): Missouri 17, Minnesota 21 (extreme)
Gary Pinkel (Missouri): Missouri 13, Minnesota NR

Holiday Bowl: USC won 45-42.
Bo Pelini (Nebraska): Nebraska 20, USC 25
Steve Sarkisian (USC): USC 19 (extreme), Nebraska NR

Russell Athletic Bowl: Clemson won 40-6.
Bob Stoops (Oklahoma): Clemson 18, Oklahoma 22
Dabo Swinney (Clemson): Clemson 15, Oklahoma 25

Military Bowl: Virginia Tech won 33-17.
Frank Beamer (Virginia Tech): Cincinnati 25, VT NR
Tommy Tuberville (Cincinnati): Cincinnati 20 (extreme), VT NR

Miami Beach Bowl: Memphis won 55-48 in triple-overtime.
Justin Fuente (Memphis): Memphis 22 (extreme), BYU NR
Bronco Mendenhall (BYU): Memphis NR, BYU NR

Belk Bowl: UGA won 37-14.
Mark Richt (Georgia): UGA 12, Louisville 20

Sun Bowl: ASU won 36-31. Cutcliffe said at the press conference he thought ASU was a top-10 team and Duke top-25.
David Cutcliffe (Duke): ASU 17, Duke 22

Boca Raton Bowl: Marshall won 52-23.
Rod Carey (Northern Illinois): NIU 20 (extreme), Marshall 25

Bitcoin St. Petersburg Bowl: NC State won 34-27.
George O’Leary (UCF): UCF 21 (extreme), NC State NR

Wednesday, December 24, 2014

Final 2014 MLB payroll shows Beltway boom; D-backs & Rangers collapse

Awhile back, I conducted research on MLB opening day payrolls and the team winning that season. I found that salaries and winning in the regular season were loosely correlated, but even a mathematical correlation doesn't necessarily mean there is one (association as causation).

In the prior posts, I used opening day payroll as my X-variable. However, now that the season is over, we have the official ending payrolls of teams. I was able to find values for ending payroll from 2000-now, using three different sources. These values are used for luxury tax purposes and accounts for things like salary dumps and midseason acquisitions, along with the rest of the 40-man roster.

The expected win percent is based purely on payroll figures from 2000-14, adjusted for inflation in salaries. From the 2014 season, the Orioles (+9.8%), Nationals (+8%), Angels (+7.8%), Pirates (+7%), and Royals (+6.4%) were the top performing teams relative to their respective ending payrolls, while the Rangers (-10.2%), D-backs (-9.7%), Red Sox (-9.1%), Phillies (-8.8%) and Rockies (-7.8%) all severely underperformed. 

TeamFinal 2014 payrollWin %ExpectedDifference
Orioles$112,707,105 59.3%49.5%9.8%
Nationals$141,803,228 59.3%51.3%8.0%
Angels$164,059,717 60.5%52.7%7.8%
Pirates$78,379,602 54.3%47.3%7.0%
Royals$97,747,983 54.9%48.5%6.4%
A's$91,615,851 54.3%48.2%6.2%
Cardinals$121,003,590 55.6%50.0%5.6%
Indians$83,697,546 52.5%47.7%4.8%
Mariners$108,957,206 53.7%49.2%4.5%
Tigers$173,291,085 55.6%53.2%2.3%
Marlins$52,518,799 47.5%45.7%1.8%
Giants$165,138,449 54.3%52.7%1.6%
Brewers$110,299,643 50.6%49.3%1.3%
Mets$92,856,260 48.8%48.2%0.5%
Blue Jays$135,435,701 51.2%50.9%0.3%
Rays$77,085,054 47.5%47.3%0.3%
Padres$85,467,063 47.5%47.8%-0.3%
Dodgers$257,283,410 58.0%58.5%-0.4%
Braves$114,699,457 48.8%49.6%-0.8%
Astros$54,689,189 43.2%45.9%-2.7%
Reds$115,358,029 46.9%49.6%-2.7%
White Sox$92,472,106 45.1%48.2%-3.2%
Cubs$93,196,617 45.1%48.3%-3.2%
Yankees$218,457,904 51.9%56.1%-4.2%
Twins$91,071,286 43.2%48.1%-4.9%
Rockies$97,975,929 40.7%48.6%-7.8%
Phillies$183,456,686 45.1%53.9%-8.8%
Red Sox$168,178,367 43.8%52.9%-9.1%
D-backs$108,124,871 39.5%49.2%-9.7%
Rangers$145,707,196 41.4%51.5%-10.2%

As you'll notice in the chart below, the Orioles, Royals, and Pirates historically underachieve. So maybe there was market correction in 2014 when those three were in the top five?

To account for the change in salaries over time, I made the CPI variable which tracked movement of average salaries, which more than doubled from the observed time period (2000-14). Basically, all the prior year salaries are converted into 2014 dollars and the multiplier factor depends on the year. Just because a team spent more in 2014, it doesn't mean the team spent more relative to the rest of the league.

TeamAvg. PayrollAvg. Baseball CPIWin %ExpectedDifference
A's$60,544,004 $83,567,393 54.5%47.7%6.9%
Cardinals$98,614,830 $137,731,557 56.2%51.0%5.1%
Braves$93,834,408 $134,321,294 55.2%50.8%4.4%
Angels$110,794,460 $149,960,153 54.8%51.8%3.0%
Giants$100,745,839 $137,427,530 53.2%51.0%2.2%
Twins$68,242,531 $91,232,713 50.3%48.1%2.1%
Marlins$45,570,320 $63,841,758 48.2%46.4%1.8%
White Sox$88,480,236 $120,549,965 51.2%50.0%1.2%
Indians$68,402,902 $98,310,102 49.8%48.6%1.2%
Phillies$114,403,870 $152,476,630 52.6%51.9%0.6%
Red Sox$139,499,156 $192,832,987 55.0%54.5%0.5%
Reds$71,297,800 $96,345,542 48.7%48.5%0.3%
Rays$50,316,371 $70,050,346 46.9%46.8%0.1%
Blue Jays$81,281,310 $111,203,841 49.4%49.4%0.1%
Padres$59,493,766 $84,221,474 47.6%47.7%-0.1%
Dodgers$129,173,887 $176,603,329 53.4%53.4%-0.1%
Yankees$193,117,372 $266,992,063 58.6%59.1%-0.5%
D-backs$79,074,909 $113,736,684 49.0%49.5%-0.6%
Rangers$93,331,986 $129,456,919 49.9%50.5%-0.6%
Brewers$70,293,258 $94,151,612 47.5%48.3%-0.9%
Nationals$66,072,385 $87,443,184 46.9%47.9%-1.0%
Mariners$91,926,390 $129,143,148 49.3%50.5%-1.2%
Rockies$72,909,516 $101,887,560 46.4%48.8%-2.4%
Astros$76,821,338 $110,186,206 46.9%49.3%-2.4%
Tigers$102,008,140 $135,648,943 48.4%50.9%-2.5%
Pirates$50,339,521 $69,588,533 44.1%46.8%-2.6%
Mets$111,941,990 $158,703,919 49.3%52.3%-3.0%
Cubs$102,876,289 $142,688,975 47.5%51.3%-3.8%
Orioles$80,747,533 $113,373,868 45.6%49.5%-3.9%
Royals$58,585,331 $79,056,701 43.4%47.4%-3.9%

Going into this research, I was expecting teams to have vastly different win% expectations using ending payroll instead of beginning. However, most teams were expected to win about the same amount of games. On average the two models differed by 0.16% (about one-fourth of a win per season). 

The A's (+6.9%) are still the highest performing team relative to payroll, while the Cardinals (+5.1%), Braves (+4.4%) and Angels (+3%) round out the top four. Although in a slightly different order, the bottom five remained the same from the last model. The Royals (-3.95%), Orioles (-3.93%), Cubs (-3.8%), Mets (-3%) and Pirates (-2.6%) all have not performed relative to how much each club spent on payroll, but as I mentioned earlier, three of them shined last year. 

Saturday, December 13, 2014

USA Today/Amway Coaches' poll ballots show familiarity bias

At the end of each season, the coaches participating in the Amway/USA Today Coaches' Poll are forced to reveal their ballot. Whether or not the coaches are actually filling them out or someone else in the athletics department is doing the work is another story. Last Sunday, we saw the ballots from coaches and there were obvious signs of the familiarity bias. Coaches were generally more favorable toward their own team than their peers in addition to giving preferential treatment for conference members or out-of-conference foes.

I ran a few tests to find out who the most extreme voters were in the Coaches' poll, using the same methodology as I did last month with AP voters. Here were the top five most extreme coaches from the final ballot:

1. Rick Stockstill (Middle Tennessee): Stockstill is a prime example of someone who ranked teams higher that his team faced. Nobody has BYU (lost 27-7), Memphis (lost 36-17), or Marshall (lost 49-14) higher than he does (No. 20, No. 16 and No. 14 respectively). He gave BYU all six of its points in the poll and Memphis 10 of its 18.

2. Frank Beamer (Virginia Tech): Beamer has the most amount of "extreme" teams among those who voted in the poll. He has Ohio State No. 2 in his poll, but the Buckeyes were only No. 2 by default when VT came to town. Beamer's most extreme ranking, though, is having Duke No. 15 (VT beat Duke 17-16). The next highest vote for Duke in the poll was No. 21.

3. Bret Bielema (Arkansas): Bielema was apparently not impressed by the teams that lost in the last week of the season. Nobody has Georgia Tech (19) and Kansas State (18) lower than him and only one person has Arizona (17) lower than he put them.

4. Art Briles (Baylor): Briles is obviously upset with the Big 12 that it promoted TCU and Baylor as co-champs, even though Baylor is the #OneTrueChampion of the Big 12. Any two-team tiebreaker would be settled by head-to-head. He ended up putting the Bears No. 3 (extreme) and TCU No. 5, with Alabama sandwiched between them.  Briles' vote for Alabama at No. 4 was the lowest the Crimson Tide received.

5. George O'Leary (UCF): Well, somebody's got to believe in Central Florida. The Knights started 0-2 with losses to Penn State and Missouri, but finished a respectful 9-1. However, that one loss is to Connecticut (2-10), which is about as ugly as a loss can get. O'Leary had UCF No. 21, and the only other vote for UCF was a No. 25 one. O'Leary also has Auburn unranked, which three others do as well. 

Other notable rankings/exclusions

-Jimbo Fisher not ranking Georgia, putting Clemson No. 12. I'm just going to assume Fisher forgot about Georgia rather than deliberately leaving them off his ballot. Georgia and Clemson are both 9-3 and the Bulldogs won the head-to-head matchup, so it wouldn't make sense for there to be that much of a difference in the other direction between the schools.

-Mike Gundy ranking TCU No. 1. Mike Gundy was really pulling for the Big 12 co-champs here. The week before TCU didn't have a first-place vote, so Gundy obviously switched his vote. Only a few people have the Horned Frogs in the Top 3, and Gundy's the only person to have them in the Top 2. He also ranked Baylor No. 3 and Oregon No. 5. Outside of his bizarre Top 5, his ballot is actually fairly closely aligned with the poll.

-Ron Turner (FIU) ranking Stanford No. 19. Stanford (7-5) certainly disappointed this season, given the Cardinal's BCS bowl streak, but it could have been better (three losses by a field goal). Turner was the only person to rank Stanford in the final coaches' ballot, and he had them at No. 19. Perhaps he ranked them because of Stanford's last game, a three touchdown wallop over UCLA, but Stanford had two less points in the poll this week than the one prior.

-Apparently, we all decided Michigan State was either the No. 7 or No. 8 team in the country. Only one voter had them different (at No. 10). Of the Top 25, the Spartans had the most conformity among voters. Wisconsin and Missouri had the least.

Conference bias

I was able to produce a ranking of each team by conference fairly easily using Microsoft Excel. For the 36 non-independent teams receiving votes in the poll, 33 of them were positioned higher in the polls by their own conference (this includes coaches voting for themselves). The three teams that were viewed less favorable were Stanford (single vote outside conference), UCLA (10.84 average vs. 10.83 Pac-12) and Wisconsin. 25 of those 33 teams were viewed as most favorable by the league they play in.

List of Extreme Teams

CoachSchoolRatingExact Coaches Matches# of extremesExtreme #1#2#3#4#5#6#7
Rick StockstillMiddle Tennessee1.31044BYU (20)Memphis (16)Marshall (14)Miss. St. (11)
Frank BeamerVirginia Tech1.13547Duke (15)Miss St (14)Ohio State (24)ASU (21)Nebraska (17)Alabama (3)UCLA (20)
Bret BielemaArkansas1.05155Georgia Tech (19)Kansas State (18)Nebraska (16)Arizona (17)Louisville (14)
Art BrilesBaylor0.96516Alabama (4)Georgia Tech (17)Louisville (12)Baylor (3)Georgia (20)Auburn (NR)
George O'LearyCentral Florida0.94743UCF (21)Arizona (18)Auburn (NR)
Mike MacIntyreColorado0.93854Air Force (24)Duke (21)Louisville (NR)Auburn (NR)
Tommy TubervilleCincinnati0.93363Cincinnati (20)Georgia Tech (16)Missouri (22)
Blake AndersonArkansas State0.91143Ole Miss (25)Northern Illinois (19)Auburn (9)
Ron TurnerFlorida International0.90574Stanford (19)Wisconsin (NR)USC (21)UCLA (20)
Les MilesLSU0.85733ASU (23)Arizona (17)LSU (18)
Troy CalhounAir Force0.80951Air Force (22)
Gary PinkelMissouri0.80074Michigan State (10)Kansas State (18)Wisconsin (8)Miss. St. (11)
Norm ChowHawaii0.79174Marshall (17)Colorado State (24)Missouri (NR)Wisconsin (NR)
Jimbo FisherFlorida State0.78884Georgia (NR)Northern Illinois (22)Louisville (13)Clemson (12)
Steve SpurrierSouth Carolina0.78582Colorado State (23)Louisville (NR)
Todd MonkenSouthern Miss.0.78052Nebraska (16)Georgia (20)
Al GoldenMiami0.77150
Dan McCarneyNorth Texas0.74770
Todd BerryLousiana - Monroe0.74752Baylor (3)Florida State (5)
Rocky LongSan Diego State0.74361Utah (19)
Craig BohlWyoming0.73341Boise State (10)
Terry BowdenAkron0.73231Oklahoma (16)
Kevin WilsonIndiana0.72263Clemson (NR)ASU (9)LSU (18)
Justin FuenteMemphis0.71052Memphis (22)Auburn (NR)
Bobby HauckUNLV0.70351Colorado State (24)
Rod CareyNorthern Illinois0.70342Northern Illinois (20)Louisville (NR)
Charlie StrongTexas0.69961TCU (3)
David BailiffRice0.68861Cincinnati (23)
Bo PeliniNebraska0.671102Minnesota (19)Louisville (NR)
Brady HokeMichigan0.66230
Mike GundyOklahoma State0.65493TCU (1)Oregon (5)Baylor (3)
Chris PetersenWashington0.65483UCLA (21)Oklahoma (19)Missouri (22)
Dino BabersBowling Green0.64871Utah (21)
Urban MeyerOhio State0.635100
Matt RhuleTemple0.63370
Nick SabanAlabama0.62881LSU (18)
Steve SarkisianUSC0.60562USC (19)Utah (20)
Frank SolichOhio0.59562Minnesota (21)UCLA (20)
Mike RileyOregon State0.59341Alabama (3)
Gary PattersonTCU0.584112Oklahoma (18)TCU (3)
Bronco MendenhallBYU0.58390
Tim DeRuyterFresno State0.57531Alabama (3)
Dennis FranchioneTexas State0.57591Mississippi State (11)
Larry BlakeneyTroy0.56972Missouri (24)Minnesota (21)
Jeff QuinnBuffalo0.56790
Bill BlankenshipTulsa0.55580
Larry FedoraNorth Carolina0.54671Mississippi State (12)
Bob StoopsOklahoma0.545111TCU (3)
Dabo SwinneyClemson0.53760
Matt CampbellToledo0.51970
Larry CokerUTSA0.51970
Jerry KillMinnesota0.51961Minnesota (21)
Paul JohnsonGeorgia Tech0.51860
Brian KellyNotre Dame0.510101Alabama (3)
Mark RichtGeorgia0.49390
David CutcliffeDuke0.49160
Rich RodriguezArizona0.47590
Mark DantonioMichigan State0.475110
Mike LeachWashington State0.46980
Bob DiacoConnecticut0.45790
Ken NiumataloloNavy0.417100

(The rating is a formula I came up with to determine how extreme a voter is. The exact coaches matches means that a coach ranked the team the same as the final poll turned out. I blued extreme teams that were from the head coach of that team.)