Sunday Dec 28, 2008

NFL Playoff Predictions

With the end of the regular season, it is time for my computer to make some playoff predictions.

First, a look back.  For the season, my computer went 143-105-5 picking against the spread and 166-86-1 picking winners.  For comparison I had the Vegas Line going 159-82-1 picking winners.  So, I had a reasonably good year.  And as far as the right teams making the playoffs, the highest ranked team that didn't make it was New England at #9 and New Orleans at #12.  The lowest ranked team that made the playoffs was Arizona at #19.

Now, for the picks and predictions.  For wildcard weekend the picks are:

  • Baltimore by 2.8 over Miami - Even though they are on the road, the computer likes Baltimore better.
  • San Diego by 0.6 over Indianapolis - Two teams playing well, but San Diego being at home makes the difference over the higher ranked Indy.
  • Minnesota by 1.2 over Philadelphia - Again, home field gives Minnesota the nod over the higher ranked Philly.
  • Atlanta by 0.8 over Arizona - Home field not enough to make up the gap for Arizona.

This is the first year I recall all 4 road teams being ranked higher.  Interesting.


As far as full projections for the playoffs, here are the chances of each team making it to and winning the Super Bowl.

Team Get to Win
Tennessee 0.366 0.197
Pittsburgh 0.354 0.194
New York Giants 0.402 0.193
Carolina 0.301 0.136
Baltimore 0.087 0.048
Indianapolis 0.083 0.044
Philadelphia 0.082 0.038
San Diego 0.064 0.035
Minnesota 0.082 0.034
Atlanta 0.074 0.031
Miami 0.049 0.030
Arizona 0.059 0.021

These predictions to take into account that 8 teams have to play wildcard weekend as well as the ratings of each team and who they are playing.

It is no surprise then that the 4 teams with byes are at the top of the list.  Of the non-bye teams, my computer likes Baltimore the best actually moving them from #2 to #1 in the rankings this week.  Of course teams like Tennessee and the Giants had nothing to play for and played a bunch of backups so a small part of any drop they had may not be fully justified.

Should be a good playoffs!


Tuesday Dec 23, 2008

College Bowl Predictions 12/23 Update

I posted my computers bowl game predictions a few days ago and with 5 games played, I thought I'd give an update.

In picking winners I went 3-2 getting Wake Forest over Navy, South Florida over Memphis, and Arizona over BYU, but missed the close Colorado State win over Fresno State and close Southern Miss win over Troy. Not a bad start as the misses were expected to be and were close games and Vegas also went 3-2.

As far as the conferences go, it is very early but here is the status:

Conference Predicted Record Actual Record
ACC 7-3 1-0
Big-Televen 1-6 0-0
Big-12 7-0 0-0
Big-East 3-3 1-0
C-USA 2-4 1-1
Independents 0-2 0-1
MAC 3-2 0-0
Mountain West 1-4 1-1
Pac-10 4-1 1-0
SEC 2-6 0-0
Sun Belt 1-1 0-1
WAC 3-2 0-1

The only conference where things are off to a bit of a wrong start is the WAC but still 4 games to go.

As far as the original predictions and any changes, apart from a few tenths here and there, the only change in a pick is to change Hawaii in a virtual tie to Notre Dame by 0.3.

There is a great game tonight with 12-0 Boise State taking on 10-2 TCU, TCU's two loses to Oklahoma and Utah. Current pick is to take the 3 points Boise State is being given as I pick them by 0.4. Enjoy!

Saturday Dec 20, 2008

College Bowl Predictions; Big-12 7-0, SEC 2-6!

With the bowl season about to start, it is time to list my computers predictions. I list one decimal point in the predicted margin for those that might need that precision for entertainment purposes. :)

Favorite Margin Underdog
Wake Forest 2.4 Navy
Fresno State 1.9 Colorado State
South Florida 8.4 Memphis
Arizona 1.1 BYU
Troy 2.1 Southern Miss
Boise State 0.3 TCU
Hawaii 0.1 Notre Dame
Central Michigan 4.7 Florida Atlantic
North Carolina 3.1 West Virginia
Florida State 6 Wisconsin
California 1.7 Miami-Florida
Northern Illinois 1.8 Louisiana Tech
Rutgers 2.5 NC State
Missouri 8.4 Northwestern
Maryland 0.1 Nevada
Rice 0.1 Western Michigan
Oklahoma State 5.2 Oregon
Air Force 0.7 Houston
Oregon State 1 Pittsburgh
Boston College 5.1 Vanderbilt
Kansas 7.4 Minnesota
Georgia Tech 5.9 LSU
Iowa 3.9 South Carolina
Georgia 1.9 Michigan State
Nebraska 1.2 Clemson
USC 0.3 Penn State
Virginia Tech 2.6 Cincinnati
Texas Tech 10.7 Ole Miss
East Carolina 2.6 Kentucky
Alabama 3.5 Utah
Connecticut 2.0 Buffalo
Texas 10.3 Ohio State
Ball State 4.1 Tulsa
Oklahoma 5.0 Florida

Now, as games are played and my ratings are updated, these predictions may change a little so I'll probably update them periodically so stay tuned.

Clearly my computer likes the Big-12 picking OU, UT, TT, and Ok. St. to all win their games, but here is how it predicts the conferences to do. Let's see how it does!

Conference Predicted Record
ACC 7-3
Big-Televen 1-6
Big-12 7-0
Big-East 3-3
C-USA 2-4
Independents 0-2
MAC 3-2
Mountain West 1-4
Pac-10 4-1
SEC 2-6
Sun Belt 1-1
WAC 3-2

Now, with this few games the specific matchups can skew things, but it is no surprise the Big-12 is expected to have the best record, although perhaps 7-0 is a surprise, but the Pac-10, considered to be having a down year is expected to go 4-1 and a lower conference the MAC is even 3-2, but the SEC is predicted to go 2-6. Even with the matchups perhaps dictating some of it, this is supported by my computer not thinking as highly of the SEC and specifically Alabama as the pollsters all year and backed up by the SEC championship game.

Enjoy the games!

Saturday Dec 06, 2008

BCS Update and New Ratings

My computers ratings are now posted and continue to have OU, Texas, and Florida #1 thru #3. Florida makes up a little ground but not enough to pass Texas.

And this is an indication of why the BCS computers will keep Florida behind Texas and create a gap that the polls will have to make up to make what the media is telling us will happen (OU vs Florida) actually do so. Jeff Sagarin's ratings are already out and Florida is #4 (!) behind Texas Tech, I've computed my best attempt at the Colley Matrix (he is kind enough to publish his algorithm) and Florida is #3 behind Texas, and I'm guessing Florida will be #3 in Anderson and Hester, #2 in Billingsley, #4 in Massey, and #4 in Wolfe. This will leave a gap of 0.04 that the pollsters must make up.

My computer did well predicting games going 12-8 against the spread overall and 2-1 in best picks. Ironically, my computer picked the Florida over Alabama "upset" but since Vegas felt even stronger about it (10 points) than my computer (6 points) it missed that game with the 11 point margin of victory. But I'll take having Alabama more or less properly ranked all season and going 8-3 picking them against the spread over picking one game right.

Stay tuned for bowl picks coming soon

Alabama was a fraud, and the top-2 in the BCS will be?

I wrote several blog entries last week first predicting OU would move to #2 in the BCS and then asking if the BCS should drop the polls where I also listed some of my computer's predictions for this weekend.

I'm pleased to say the computer picked the "upset" by Florida over Alabama as I wrote about a month ago saying Alabama was a fraud (and Alabama lost by more than I predicted so my computer having them #7 may have been high!) and now the question is who will be in the top-2 of the BCS come Sunday evening? The media would have us believe that OU vs Florida is a foregone conclusion, but I'm not so sure.

Coming into this week, Florida was #4 in the BCS trailing #3 Texas by 0.0372 which trailed #2 OU by 0.0128. In the polls, the 3 teams were virtually identical with OU totaling 1.825, Texas 1.827, and Florida 1.835, the deciding factor being the computers where OU was at 0.980, Texas 0.940, and Florida trailing at 0.820. With Florida beating Alabama, they will improve in the computer ratings but my guess is that they will still be #3 behind OU at 1.000, Texas at 0.940, and Florida at 0.900.

To make up the gap in the computers, Florida will have to be a combined total of at least 175 votes ahead of Texas in the polls. This is 142 votes more than the gap last week. For this to happen, voters will have to either a) in dropping Alabama, give many more of their #1 votes to Florida than to OU, or b) those that had Florida behind Texas last week will have to swap them.

Correction: Given Florida was already #2 in the Harris poll ahead of Texas but #4 behind Texas in the Coaches poll, I'm estimating they'll need to extend the lead in the Harris poll to 88 points and take a lead of 14 points in the Coaches poll to overcome the deficit in the computer ratings.

While one or some combination of both may happen, it will be because voters are manipulating the system to get the championship game they want. I fail to see the logic in why Florida should get more of Alabama's #1 votes than OU given OU's dominating win over Missouri, and if last week a voter had Texas ahead of Florida and the team Texas beat just walloped Missouri, why does Florida beating an overrated Alabama justify changing that?

Another way of looking at it is that we have 3 1-loss teams so lets look at which one has the best or worst loss. OU lost to Texas so that would seem to be the "best" loss which confirms them being #1. Texas' loss was to a 1-loss Texas Tech which would seem to be a better loss than Florida's loss to a 4-loss Ole Miss. By that logic USC should be being considered as their loss was to a 4-loss Oregon State.

So the voters will likely vote to make it happen because the media says it should happen, but watch closely to see how close it is and don't be surprised if honest voting ends up having OU and Texas in the BCS championship game.

Sunday Nov 30, 2008

Should the BCS drop the polls?

I thought it would be interesting to take a summary look at the BCS, polls, and some computer ratings to see how they compare. Below, is such a table with the first number in each cell the current ranking and the second number from 4 weeks ago to see how things have changed.

Texas Tech7/28/38/38/24/T15/4
Penn State8/36/26/26/39/46/5
Boise State9/109/109/109/97/89/10
Ohio State10/1110/1210/1110/1211/1211/11

The first observation is that the three polls, while independent and made up of different voters, end up with near identical rankings. This either means that they must be right because they agree, or they agree because they influence each other and the voters use others as input when submitting their votes and they aren't necessarily correct and aren't independent. I suspect it is more of the latter. If I'm right, then you have ask if using two polls it the right way to determine 2/3 of the BCS? If they just follow each other what is the point?

The second observation is that the polls and BCS computers vary quite a bit from week 10 to week 14 (and thus the BCS varies too since it is computed from these). You would think that a ranking system should do a pretty good job of ranking teams where they should be and by week 10 should be pretty accurate and thus a team shouldn't change a bunch by week 14 if the ranking system had things relatively correct in week 10. But looking at the RMS deviation for these you see a range of 2.55 (Harris) to 3.38 (BCS Computers). However, if you take a look at my computers rankings, the RMS deviation is only 2.1. Now if a ranking is just stubborn and doesn't move teams appropriately it could have a small RMS deviation but not be accurate. This leads to the third observation.

The third observation is that the computers, and particularly my computer don't agree that Alabama is the best team in the country like the polls do. So are they correct in doing so? At least my computer has done very well in predicting Alabama games going 8-2 predicting their games against the spread so that would seem to support the ranking. Of course, we have another chance to test this theory with Alabama playing Florida this weekend where my computer has Florida as a nearly 6 point favorite. And my computer is being conservative as Vegas has Florida a 10 point favorite!

So, from the first observation we see the polls aren't very independent and perhaps not objective and that there is little point in using multiple. From the second observation we see that the polls don't tend to have things right and either have to adjust to get them right, or overreact and thus shouldn't be trusted. From the third observation we see that the computers, particularly mine address the first two issues, so perhaps we should throw out the polls and use computer rankings alone!

Not that one week should be the deciding factor (and clearly my computer has picked Alabama games well this year over many weeks), but here are my computers picks for the big games this weekend:

  • Ball State by 11.4 over Buffalo
  • Navy by 11.6 over Army
  • USC by 22.1 over UCLA
  • OU by 12.7 over Missouri
  • Florida by 5.8 over Alabama
  • Boston College by 1.7 over Virginia Tech

It should be fun to watch!

Oklahoma to move to #2 in BCS over Texas

With just one computer publishing its ratings (Sagarin) and my estimating another (Marsee), OU has already moved ahead of Texas in the BCS ratings. As the rest of the computers report, the gap should only grow meaning the voters will have to have a pretty big "change of mind" and swap Texas ahead of OU to keep Texas #2.

So it looks like it will be OU against Missouri in the Big-12 championship and Florida against Alabama in the SEC. If OU wins, they are in the championship game and likewise if Alabama wins they are in. Florida winning should also get them there even though they'll likely be behind Texas in this weeks rankings as beating Alabama should vault them ahead. I suppose there is an outside shot at an OU vs Texas rematch but it is highly unlikely. If Missouri somehow won, it possibly opens the door for USC since they would be a conference champ and Texas wouldn't be, but more likely Texas would get the nod.

So, the national championship game is likely one of:

  • OU vs Florida
  • OU vs Alabama
  • OU vs Texas (unlikely)
  • Texas vs Florida
  • Texas vs Alabama
  • USC vs Florida (very very unlikely)
  • USC vs Alabama (very very unlikely)

Saturday Nov 29, 2008

Boise State in the BCS?

As the College Football season winds down, the discussion becomes who will play in the BCS bowl games, and this year specifically with 3 non-BCS team being undefeated, which of those will "bust" the BCS.

The popular pick this year is Utah to be a BCS buster given they have been the highest rated team of the 3 non-BCS teams (Boise State and Ball State the other two). My computer has actually had Boise State rated higher all year with Boise State #12 and Utah #13 coming into this weekends games. Ball State is #16. So it would seem that perhaps it shouldn't be just a given that Utah is the choice, but as I sit here watching the games, perhaps more than 1 can be chosen!

I say that because Oregon State is currently losing to Oregon. Why does that matter? Follow me here and it will all make sense

There are 5 BCS games (10 teams) with the 6 conference champions from the major conferences, the Pac-10, Big-12, Big-11 (a story for another blog post, how can an 11 team conference be called "10"), ACC, SEC, and Big-East, all guaranteed a spot. That leaves 4 at-large berths which have typically been filled with the 2nd place teams from the BCS conferences and Notre Dame. Note that a conference cannot have more than 2 teams in the BCS. Both Utah and Boise State have been able to "bust" the BCS in past years, but not in the same year.

So with Notre Dame stinking it up this year (about to go to 6-6) and 3 undefeated teams from non-BCS conferences and potentially none from BCS conferences (I predict Florida will beat Alabama next weekend) this could be the year. So let's take a look at the likely scenarios:

  • Pac-10 - Assuming they win out (just need to finish Notre Dame tonight and then a weak UCLA next week) USC are in the BCS, but might not win the Pac-10 if Oregon State beats Oregon tonight in which case Oregon State would also be in, so 2 of the 10. But if Oregon wins tonight, Oregon State would not make it and the Pac-10 only gets 1 team and this is when it gets interesting (see below).
  • Big-12 - They will get 2 teams, we just don't know which 2. Could be any of Texas, Oklahoma, Texas Tech, and Missouri.
  • Big-11 - Penn State, and that should be it but Ohio State has an outside shot (see below).
  • ACC - Arguably they shouldn't have any this year! But rules are rules and they'll get 1 from their championship game next week.
  • SEC - They will get 2 teams, Florida and Alabama, we just don't know which goes to the BCS championship game yet but that will be decided next weekend.
  • Big-East - See ACC.
  • Utah - Everyone already concedes they will get in with their season finished and BCS ranking high enough.

So you count that up, and you get 10 so it is all done right? Well, Oregon is beating Oregon State, so we'd only add up to 9 then and we have to find another at-large team. The usual pick would be to go find another team from a BCS conference and the first in this list would be Ohio State (can't take 3 Big-12 teams, see rule above), but I would argue a 2-loss Ohio State that didn't really play anyone other than Penn State and USC (their two losses) should not be going which opens the door for Boise State!

Will it happen, I certainly hope so, but doubt it. But wouldn't it be fun to have two undefeated teams (Utah and Boise State) playing each other? The logical place for this to happen would be the Fiesta bowl, but the Fiesta would have to pass up (potentially) a Texas or Oklahoma and all the fans that would follow.

Of course, by the time you read this Oregon State could come back and make this all moot. But I wanted to be the first (?) to get this blogged about so happy reading and speculating!

Saturday Nov 08, 2008

Alabama is a Fraud

Ok, perhaps not a complete fraud, but certainly not (IMHO) a deserving #1 in the college football rankings. How could an undefeated team be a fraud you ask?

To start, lets look at their schedule. Of their 10 opponents, only 4 even have winning records. This results in an opponents percentage of only 0.518 even with the losses to Alabama factored out. Further, Alabama's out of conference schedule was Clemson (4-5), Tulane (2-7), Western Kentucky (2-8), and Arkansas State (4-5).

Compare this with the opponents winning percentages for the other undefeated teams and Alabama falls behind Texas Tech (0.602) and Utah (0.576) and only ahead of Boise State (0.468) and Ball State (0.387). At least Utah went out of conference against Oregon State (6-3) and Boise State against Oregon (7-3). Alabama's opponents winning percentage also falls behind 1-loss contenders Texas (0.693), Florida (0.610), Southern California (0.539), and Oklahoma (0.539). So they clearly haven't played the toughest schedule and if someone like Southern Cal is going to get dinged for a weak schedule it would seem that Alabama should too.

But a team shouldn't be penalized for having a poor schedule if they soundly defeat the poor teams and do play well in their games against better teams. And this is where Alabama falls down a bit. In a home game, they only beat a 2-7 Tulane team by 14, they eked out a 3 point win again at home against Kentucky, and eked out another 4 point win again at home against Mississippi. Their two significant wins are against Georgia by 11 and a weaker than normal LSU by 6. And Georgia got walloped by 39 and LSU by 30 against Florida, and LSU, at home, was able to beat a 3-6 Mississippi State by only 10 and Georgia was able to beat Kentucky by only 4. So, Alabama isn't soundly defeating some not so good teams and their only 2 significant wins were not that impressive.

Compare this again with their undefeated competition and you see Texas Tech soundly defeating all their poor opponents (closest win 16, others in 20's and above) and has beat a 1-loss Texas in a close game, walloped an 8-2 Oklahoma State by 36, and a 6-4 Kansas by 42. Similarly 1-loss Oklahoma, Texas, and Southern Cal, and Florida have clearly handled poor competition and aside from their 1 loss each, good competition as well.

Now of course, those 1-loss teams have, well, 1 loss, so doesn't Alabama staying unblemished trump these other failings? To a degree yes, but clearly there is a point where it doesn't (see Utah, Boise State, and Ball State) so why should Alabama be any different? I think it is clear that based on the data one can't justifiably put Alabama ahead of Texas Tech but using the same logic as is used for not putting Utah or Boise State ahead of Oklahoma, Florida, Texas, and Southern Cal one could argue Alabama shouldn't be ahead of them either.

For an objective look at who should be ranked where, see my computer ratings which have Texas Tech #1 but some surprises after that.

Now, thankfully time will tell, especially with Alabama having to face Florida in the SEC Championship, so either my skepticism will be proven right or Alabama will have the opportunity to show they deserve a BCS championship game berth. Of course, they have two more poor/average teams to reinforce my analysis above before that (3-6 Mississippi State and 5-5 Auburn).

Friday Nov 07, 2008

Track links for 2008-11-8

Saturday Oct 18, 2008

NCAA Cross Country Heating Up

While it is generally not as thrilling or followed as the outdoor track and field season, many college track athletes are well into their cross country season with the Pre-Nationals event being held yesterday.

Held in Terre Haute, Indiana on the same course that the National Championships will be held on in a month, but over 2K shorter distances for both the men and women (8K and 6K respectively), it offers a preview of what's to come at Nationals. There are 2 races each, a blue and white, for both the men's and women's teams as well as an open race, so as to not have all the teams competing head to head just like they will at Nationals. Not all of the schools show up for the event either so there will still be some surprises when they line up for the championship in a month.

The highlight of the event was Liberty's Sam Chelanga setting a new course record of 22:51.3, besting former teammate Josh McDougal's record from last year, in the white race. He clearly sets himself as the favorite for the individual title and overshadowed Stanford's team win in the white race beating Iona. In the blue race Alabama held off Northern Arizona. Oklahoma State ran in the team race but held some of their runners out to run in the open race including German Fernandez, who won in 23:34.57. You will recall that German set records as a California prep this spring.

In the women's races, #1 Washington impressed totaling an impressive 36 points in the blue race with runners including Christine Babcock who herself set prep records in California this spring.

I'm looking forward to Pac-10's, regionals, and nationals over the next several week!

Sunday Oct 05, 2008

College and NFL Football Thoughts and Rankings

I am a big sports fan, as has probably been apparent from some of my blog entries, and as such am into football. I've generally been a fan of the AFC, particularly the AFC West having grown up and lived most of my life on the west coast, and as you would expect am also a Pac-10 fan.

In addition to being a fan and watching a lot of games, my inner geek gets the better of me and I dust off my computer ratings each year and have done so yet again and you can see them here. I've actually been doing them fairly regularly since 1991 when I got started after Georgia Tech's disputed championship as I thought there should be a way to objectively determine who is #1. I used Jeff Sagarin's ratings as my inspiration but have worked to tune my system to both have ratings that pass the "smell test" but also ones that do well in predicting games against the spread.

Over the years, I added in doing ratings for the NFL too and have toyed with doing them for college basketball, but it is most entertaining to look at the college ratings as that is where there is no playoff and one can argue for hours on end who is #1.

The current college ratings have Oklahoma #1, Missouri #2, Penn State #3, and USC #4. Going into this weekend Dallas was #1 in the NFL despite their loss with Tennessee close behind at #2.

In addition to doing ratings and weekly predictions, my system is able to project final records. Interestingly there are six (!) teams projected to finish the college season undefeated and in the NFL Tennessee is projected to finish 14-2 ahead of a bevy of teams at 12-4. It is still early though so things can certainly change.

So, take a look and let me know what you think.

Thursday Aug 14, 2008

Track at the Olympics is starting!

Now that we've endured the opening ceremonies, swimming, gymnastics, and the U.S. basketball teams rolling over their competition, it is now time for the real events at the Olympics to start!

Ok, just kidding, the aforementioned events have been great, but track and field does start tonight with the Heptathlon, heats of the mens 100m, and heats in the womens 800m. It will be good to see what Bolt, Gay, and Powell have to offer and also see what Pamela Jelimo chooses to throw down in the first round.

Thankfully, I have CBC to watch and see things live while the rest of the west coast has to wait 3 hours for NBC's coverage.

More to come later.

Tuesday Aug 12, 2008

Track and Olympics links for 2008-8-12

Saturday Aug 09, 2008

China now censoring polution statistics

I know I've included numerous entries as part of my track and Olympic links blog entries on the pollution in Beijing, but this latest story I saw deserves its own entry.

According to, a British smog monitoring unit was forced to be shutdown. It had already been limited to registered users but when it was reporting that pollution levels in the area many Olympic events are at were above an Air Pollution Index of 100 (which is agreed upon as unsafe) while the governments measurement was 95, that was apparently too much.

I'm not sure if I'm more disappointed that the Chinese asked for the site to be shutdown or if the British group gave in. Perhaps they didn't have an option ...



« June 2016