NFL Week 4

Week 4 is posted and there are what appear to be some surprises in the team rankings for this week.

A team’s record does not necessarily reflect its ability. Take the Cleveland Browns and their 1 – 3 record. It would appear that they are not all that good of a team. I recall hearing a few analysts expressing surprise that they won last weekend as well. Take a look at the results of their games so far and you’ll see why they are the 11th best team in this weeks rankings.

They lost to Tampa Bay 17-14, Kansas City 16-14, and Baltimore 24-17. KC and Baltimore are two of the top three rated teams in the league. It is no surprise to me that they defeated Cincinnati 23-20. The cumulative difference in their game scores is a total of minus nine points. Their next five games will tell a lot about their ability. All five of those opponents are currently 3 -1 (Atlanta, @Pittsburgh, @New Orleans, New England and the Jets).

On the other side of the ledger, Super Bowl Champion New Orleans is 3 – 1 and is near the bottom of the ratings because of their apparently weak schedule to date. They’ve beaten Minnesota 14 – 9, an anemic offensive team in San Francisco 25-22, and a horrible Carolina team, 16-14. They lost to Atlanta 27-24. So, while their cumulative score difference is similar to Cleveland (+7), but on the positive side, their victories are over opponents which are bad teams and teams they should have beaten badly and they lost to a nice team in Atlanta (3 – 1).

That said, I suspect a sure climb in the rankings from New Orleans and am unsure of what to expect from Cleveland.

College football 2010

College football season is almost upon us! With that, we launch into another season full of debate about the accuracy of the ratings of the Bowl Championship Series. My methodology is the only known scientific method used to measure the ability of all 730 collegiate football teams. That includes all teams from the BCS schools through the NAIA and NCAA Division III. My simple method, using the Rasch Model, objectively measures team ability using the whole data matrix. In other words, instead of using only a team’s schedule, that of their opponents and that of their opponents’ opponents, every single game is used to calculate every team’s ability measure. I do this for every competitive level of college football.

Due to the sparse nature of the data matrix (only 9% of the matrix is populated, leaving the other 91% void of any data) I have to group the teams into two separate groups instead of all teams being measured using the same ruler. Still though, BCS and FCS schools are grouped together. These are the teams that generate the most interest. NCAA Division II, III and NAIA are in the second group. These groupings allow for actual comparison of teams of the same group, not just within the same collegiate competitive division.

My method not only measures the ability of teams, it also shows the standard error of measurement so it can be seen how expected or unexpected a particular game’s result was or the probability of victory in the next game. Unpublished is a diagnostic statistic (infit/outfit) which tells the reader how unexpected the team performed over the course of the season. This provides very good information on how consistent a team performed.

This method is also easily duplicated for every sport at every level – high school through the pros. This method brings fairness to all sports in that it measures team ability without the subjective input from humans (i.e. voters polls). The rightfully deserving teams make the playoffs with a chance for competing for the championship of their particular division. My method is easily instituted for any sport on any level.

Subscribe to this blog, my twitter account (@SportsMeasures) or, my Facebook page: http://www.facebook.com/pages/Sports-Performance-Measurement/121166336628

Please feel free to share your thoughts here about any of the topics presented or, even start your own thread about sports performance measurement!

Masters Golf

Check out the SportsMeasures.com blog later this week for complete analysis of the 2010 Masters golf tournament! You will see golf analysis unlike anything you’ve seen before. Starting Wednesday I will explain some of the details that this analysis will cover: hole difficulty, unexpected scores (by hole), player ability for example. I hope you post your thoughts and reactions to this analysis.

Sweet 16!

WOW! What an opening weekend to the tournament! There is no measurement methodology available that could have accurately predicted these games. Someone with some lucky guesses, sure. But there is no science available that could have seen this coming. Science aside though, there were some great games.

Looking to the second weekend though, we have some compelling match-ups on Thursday and Friday. I still like St. Mary’s to continue on their way over Baylor and then face off against Duke for the regional championship. I LOVE St. Mary’s style of play and they are very skilled at it.

My Midwest bracket is totally blown up – I have one team remaining in Ohio St. Two teams I really liked coming into the tourney – Kansas and Georgetown – really shocked everyone. Kansas’ measure was the best by nearly one and one-half standard errors over Kentucky. That means that they would have been expected to beat even the #5 team (West Virginia) two out of three meetings. They were prohibitive favorites against Northern Iowa with a difference of just over 12 logits in their measures! Kansas would have been expected to beat UNI nearly every single time they played. UNI found that one time out of more than 1000 that would take to beat a team with a measure that much greater than their own. This was an extremely unexpected result. I wasn’t going to analyze the tournament games but, with this upset, I would like to see the model’s degree of unexpectedness. I will discuss that in another post.

Georgetown surprised me greatly too. They didn’t just get beat by Ohio, Ohio beat them soundly and scored 99 points on top of that. Georgetown was one of the better defensive teams in the country and the Big East Conference tournament performance put an exclamation point on their season by showing that they were just as good as WVa, Syracuse or Villanova. The Big East’s performance in the tournament so far has also been a big surprise. Perhaps, the Big East isn’t as good as we all thought. Of the eight teams in the tournament from the Big East, only two remain.

Regarding my bracket picks, I did my best to stick to the measures with only one exception. I really felt Siena would take down a hampered Purdue squad. Without their leading scorer in Robbie Hummel, I felt Purdue was ripe for the picking in the first round. They have proven me wrong with an exclamation point.

Looking forward to Thursday!

Lastfanstanding2010.com

For the duration of my entry into the contest at www.lastfanstanding2010.com I will be blogging there. As soon as I lose in this contest I will return to SportsMeasures.com to continue blogging about sports. – Patrick

SportsMeasures.com field of 65

Here is the field of 65 as seen by SportsMeasures.com. The columns are as follows: Seed, SM.com ranking, team, measure, standard error

1    1        Kansas    76.90    2.54
1    2        Kentucky    73.37    2.57
1    3        Syracuse    72.78    2.14
1    4        Duke    72.58    2.00
2    5        West Virginia    69.74    2.00
2    6        Kansas St    69.43    2.03
2    7        Villanova    68.98    2.07
2    8        Purdue    68.51    2.09
3    9        Brigham Young    68.41    2.12
3    10        Baylor    68.15    2.03
3    11        St Mary's CA    67.79    2.95
3    12        Texas    67.44    1.92
4    13        Ohio State    67.43    1.97
4    14        Georgetown DC    67.22    1.92
4    15        New Mexico    67.18    2.19
4    16        Maryland    67.10    2.00
5    17        Temple    66.94    2.12
5    18        Wisconsin    66.91    1.98
5    19        Texas A&M    66.41    1.99
5    20        Tennessee    66.18    2.00
6    21        Butler    66.15    2.22
6    22        Pittsburgh    65.92    1.98
6    23        Xavier OH    65.92    1.97
6    24        Vanderbilt    65.75    2.00
7    25        Missouri    65.50    1.97
7    26        California    65.45    1.88
7    27        Michigan St    65.43    1.95
7    28        Marquette    65.17    1.91
8    29        Clemson    64.92    1.93
8    30        Louisville    64.86    1.94
8    31        Northern Iowa    64.81    2.18
8    32        UTEP    64.71    2.04
9    33        Florida St    64.54    1.95
9    34        Notre Dame    64.21    1.90
9    35        Washington    64.18    1.87
9    36        UNLV    64.13    2.01
10    37        Georgia Tech    63.99    1.86
10    38        Oklahoma St    63.92    1.95
10    39        Utah St    63.61    2.01
10    40        Minnesota    63.61    1.86
11    41        Virginia Tech    63.45    1.99
11    42        San Diego St    63.35    2.05
11    43        Richmond    63.22    1.93
11    44        Old Dominion    63.01    1.91
12    45        Mississippi    63.01    2.00
12    46        Memphis    62.79    1.98
12    51        Cornell NY    62.24    2.53
12    55        Siena    61.78    2.08
13    56        Murray St    61.70    2.25
13    92        Wofford    57.15    1.96
13    96        Houston    56.72    1.86
13    98        New Mexico St    56.70    1.98
14    101        Montana    56.22    1.99
14    102        Oakland    56.15    2.09
14    103        Sam Houston St    56.15    2.11
14    109        Ohio U.    55.63    1.82
15    134        UC-Santa Barbara    53.92    2.04
15    141        Morgan St    53.23    1.93
15    146        North Texas    52.93    2.06
15    150        Vermont    52.82    1.93
16    166        East Tennessee St    51.68    1.87
16    177        Robert Morris PA    50.92    1.92
16    181        Lehigh    50.75    1.94
16    231        Arkansas-Pine Bluff    46.60    2.00

Week to week ranking movement

Kansas is still #1 this week despite its loss last week and will most likely retain the top spot in next weeks poll with its win over K-State Wednesday night!

Just because a team loses once or twice doesn’t mean that the team has lost some of its luster. Kansas is the class of the 2009-10 men’s basketball season and they *should* win the NCAA tournament championship early in April. Kansas lost to a decent Oklahoma St. team on the road. Road games in almost any conference are potential losses – even for the top team in the country. The fact that they were 14-0 in conference play up to that point is a testimony to how good this Bill Self-coached team is!

What is even worse than the scenario described above is when a team loses to a team of equal ability and still is rated lower in the following week’s rankings – even when it is a close game! A loss does not always equal lower ability and subsequent lower ranking!

This is just another piece to the argument that subjective opinions and analysis has no place in such high stakes athletic competition!

Conference strength

Here are the top six men’s basketball conferences for the 2009-10 season:

Big 12 – 64.63

ACC – 63.78

Big East – 63.57

SEC – 61.51

Big 10 – 61.11

Pac-10 – 58.70

It is no surprise that the Big 12 is at the top of the list. It is somewhat surprising to see the ACC in the #2 hole though. The Big East, with Syracuse and Villanova having outstanding seasons, it would seem, would be the second strongest conference in the nation. However, the ACC’s weakest teams still have measures above 59 (it will take a measure of about 64 or so to get into the NCAA tournament, as of now). Every other conference has multiple teams with measures below 59.  Even with the SEC having a down year, the Big 10 is still showing it is not as strong as its reputation would indicate, or the public rankings of their top teams. The Big 12 should produce this year’s national champion, most likely Kansas. However, the Big East could place two teams in the Final Four. Although, Kansas St could sneak up on someone even though they’ll likely have a two seed and perhaps even a #1. I think the teams who make the Final Four will not surprise anyone this year. After the top 10 teams, there is a serious drop off.

#2, Projected NCAA men’s basketball tournament

Projected seedings for NCAA men’s basketball tournament

Ranking,  Seed, Team name, Measure, Error
1    1    Kansas    78.83    3.24
2    1    Syracuse    75.60    2.67
3    1    Kentucky    75.35    3.31
4    1    Duke    74.32    2.24
5    2    Purdue    71.68    2.47
6    2    Villanova    71.63    2.48
7    2    Kansas St    71.02    2.41
8    2    West Virginia    70.72    2.29
9    3    Texas    70.36    2.21
10    3    Brigham Young    70.29    2.51
11    3    New Mexico    68.53    2.45
12    3    Baylor    68.12    2.28
13    4    Pittsburgh    67.99    2.22
14    4    Missouri    67.76    2.19
15    4    Maryland    67.69    2.21
16    4    Wisconsin    67.66    2.14
17    5    Georgetown DC    67.59    2.27
18    5    St Mary’s CA    67.53    3.15
19    5    Vanderbilt    67.50    2.29
20    5    Tennessee    67.44    2.28
21    6    Ohio State    67.34    2.16
22    6    Xavier OH    66.96    2.18
23    6    Texas A&M    66.87    2.24
24    6    Louisville    66.76    2.18
25    7    Clemson    66.74    2.17
26    7    Temple    66.73    2.33
27    7    Butler    66.73    2.28
28    7    Michigan St    66.35    2.10
29    8    Florida St    66.03    2.16
30    8    Virginia Tech    65.88    2.28
31    8    Northern Iowa    65.82    2.44
32    8    Marquette    65.71    2.20
33    9    California    65.38    2.08
34    9    Wake Forest    65.27    2.19
35    9    Georgia Tech    65.26    2.11
36    9    UTEP    64.95    2.32
37    10    Oklahoma St    64.72    2.19
38    10    UNLV    64.51    2.23
39    10    Gonzaga    64.16    2.36
40    10    Florida    64.06    2.15
41    11    Washington    63.97    2.03
42    11    Richmond    63.94    2.17
43    11    Memphis    63.94    2.28
44    11    Connecticut    63.94    2.08
45    12    Mississippi    63.77    2.21
46    12    Alabama-Birmingham    63.72    2.26
49    12    Utah St    63.56    2.25
52    12    Cornell NY    63.07    2.70
53    13    Old Dominion    62.99    2.06
60    13    Murray St    62.08    2.59
62    13    Siena    61.63    2.22
88    13    Kent St    58.38    2.11
101    14    Weber St    56.91    2.24
102    14    Wofford    56.61    2.11
108    14    Sam Houston St    55.94    2.40
118    14    Oakland    55.34    2.23
122    15    Coastal Carolina    55.00    2.37
135    15    UC-Santa Barbara    53.97    2.24
152    15    Morgan St    52.48    2.10
168    15    North Texas    51.81    2.18
171    16    Stony Brook    51.52    2.25
174    16    Jacksonville FL    51.32    2.10
187    16    Robert Morris PA    50.18    2.08
206    16    Lehigh    48.67    2.08
259    16    Jackson St    44.04    2.27

Projected NCAA men’s basketball tournament

To create my projected NCAA Men’s Basketball Tournament field of 65 teams, the conference leader as of today was each conference’s “champion” and received the automatic bid (New Mexico has a .5 game lead over BYU but, BYU has a better measure. There may be other instances where the conference leader is not the best team according to the measures produced by SportsMeasures.com). The at-large bids were then given to the next 34 teams in measure order. No other consideration was given: not to win/loss record, conference standing, RPI, strength of schedule, etc.

It is the contention of SportsMeasures.com that the “play-in” game should be played between the last two teams to receive at-large bids. The two worst teams, Jackson St and Lehigh, at least won their conference and should be rewarded as such and allowed to play as the #16 seed. In this field of 65, Mississippi and UAB would have to play in the play-in game. You can see the last four in at the bottom of the list. The first four out of the tournament would be Arizona St., Minnesota, Dayton, and San Diego St. ESPN has St. Mary’s as the last team in meaning they are no better than about the 45th best team in the country. SportsMeasures.com has St. Mary’s as the #18 team in the land and a solid #5 seed. ESPN would have you believe that they are barely a #10 or #11 seed. This is all because of the subjective nature of their analysis being used to determine who should be in and who should be out. The field of 65 should look more like this list than anything ESPN or anyone else is producing.

It should be reiterated that all analyses done for this site are completely scientific, reproducable, robust, and will withstand the scrutiny anyone wishes to bring. If you were to produce your own objective analysis, it would look exactly like this.

Conference Leaders as of February 27, 2010 (automatic bids)

1.     America East,     Stony Brook (#171, 51.52)
2.     ACC,     Duke (#4, 74.32)
3.     Atlantic Sun,     Jacksonville (#174, 51.32)
4.     A-10,     Temple (#26, 66.73)
5.     Big East,     Syracuse (#2, 75.60)
6.     Big 10,     Purdue (#5, 71.68)
7.     Big South,     Coastal Carolina (#122, 55.00)
8.     Big 12,     Kansas (#1, 78.83)
9.     Big Sky,     Weber St (#101, 56.91)
10.     Big West,     UCSB (#135, 53.97)
11.     Colonial Athletic,     ODU (#53, 62.99)
12.     Conference USA,     UTEP (#36, 64.95)
13.     Horizon,     Butler (#27, 66.73)
14.     Ivy,     Cornell (#52, 63.07
15.     Metro Atlantic,     Siena (#62, 61.63)
16.     Midwestern,     Kent St. (#88, 58.38)
17.     Mid-Eastern,     Morgan St. (#152, 52.48)
18.     Missouri Valley,     Northern Iowa (#31, 65.82)
19.     Northeast,     Robert Morris (#187, 50.18)
20.     Mountain West,     New Mexico (#11, 68.53)
21.     Ohio Valley,     Murray St. (#60, 62.08)
22.     Pac-10,     California (#33, 65.38)
23.     Patriot League,     Lehigh (#206, 48.67)
24.     SEC,     Kentucky (#3, 75.35)
25.     Southern,     Wofford (#102, 56.91)
26.     Southland,     Sam Houston St. (#108, 55.94)
27.     Southwestern,     Jackson St. (#259, 44.04)
28.     Summit,     Oakland ((#118, 55.34)
29.     Sun Belt,     North Texas (#168, 51.81)
30.     West Coast,      Gonzaga (#39, 64.16)
31.     Western Athletic,      Utah St. (#49, 63.56)

At-large bids for the next best 34 teams base solely on the measure

32. Villanova (#6, 71.63)
33. Kansas St. (#7, 71.02)
34. West Virginia (#8, 70.72)
35. Texas (#9, 70.36)
36. Brigham Young (#10, 70.29)
37. Baylor (#12, 68.12)
38. Pittsburgh (#13, 67.99)
39. Missouri (#14, 67.76)
40. Maryland (#15, 67.69)
41. Wisconsin (#16, 67.66)
42. Georgetown (#17, 67.59)
43. St. Mary’s (#18, 67.53)
44. Vanderbilt (#19, 67.50)
45. Tennessee (#20, 67.44)
46. Ohio St. (#21, 67.34)
47. Xavier (#22, 66.96)
48. Texas A&M (#23, 66.87)
49. Louisville (#24, 66.76)
50. Clemson (#25, 66.74)
51. Michigan St. (#28, 66.35)
52. Florida St. (#29, 66.03)
53. Virginia Tech (#30, 65.88)
54. Marquette (#32, 65.71)
55. Wake Forest (#34, 65.27)
56. Georgia Tech (#35, 65.26)
57. Oklahoma St (#37, 64.72)
58. UNLV (#38, 64.51)
59. Florida (#40, 64.06)
60. Washington (#41, 63.97)
61. Richmond (#42, 63.94)
62. Memphis (#43, 63.94)
63. Connecticut (#44, 63.94)
64. Mississippi (#45, 63.77)
65. Alabama-Birmingham (#46, 63.72)