Blog Archives

NFL Picks – Week 3

Total –  SU: 32-16-0 (66.67%) ATS: 27-20-1 (57.29% +5 Units) O/U: 31-17-0 (64.58% +12.3 Units)

Week 1 – SU: 10-6-0 ATS: 8-8-0 O/U: 13-3-0

Week 2 – SU: 10-6-0 ATS: 10-6-0 O/U: 10-6-0

Week 3 – SU: 12-4-0 ATS: 9-6-1  O/U: 8-8

Tampa Bay at Atlanta

Prediction: Falcons 24-20 (62.5%)

Pick: Buccaneers +6.5 (55.8%)

Total: Under 45

San Francisco at Arizona

Prediction: 49ers 22-20 (55.9%)

Pick: Cardinals +3 (52.7%)

Total: Under 42.5 

San Diego at Buffalo

Prediction: Bills 23-22 (51.8%)

Pick: Chargers +2.5 (55.3%)

Total: Over 44.5

Tennessee at Cincinnati

Prediction: Bengals 24-19 (64.5%)

Pick: Titans +7 (55.1%)

Total: Under 43.5 

Baltimore at Cleveland

Prediction: Ravens 22-21 (53.5%)

Pick: Browns +2 (52.2%) PUSH

Total: Over 42

Green Bay at Detroit

Prediction: Lions 25-24 (53.4%)

Pick: Lions -1 (50.5%)

Total: Under 52.5

Indianapolis at Jacksonville

Prediction: Colts 23-20 (57.4%)

 Pick: Jaguars +7 (62.3%)

Total: Under 45.5 

Kansas City at Miami

Prediction: Dolphins 21-19 (55.9%)

Pick: Chiefs +5 (58.2%)

Total: Under 41.5

Oakland at New England

Prediction: Patriots 29-19 (77.2%)

Pick: Raiders +14.5 (61.5%)

Total: Over 47

Minnesota at New Orleans

Prediction: Saints 27-20 (68.3%)

Pick: Vikings +9.5 (58.0%)

Total: Under 51

Houston at NY Giants

Prediction: Giants 22-21 (51.1%)

Pick: Giants +2.5 (58.2%)

Total: Over 42

Washington at Philadelphia

Prediction: Eagles 25-22 (59.8%)

Pick: Washington Football Team +6.5 (58.6%)

Total: Under 50

Denver at Seattle

Prediction: Seahawks 24-22 (54.6%)

Pick: Broncos +4.5 (58.2%)

Total: Under 49

Dallas at St. Louis

Prediction: Cowboys 23-22 (53.0%)

Pick: Rams +2 (52.7%)

Total: Under 45.5

Pittsburgh at Carolina

Prediction: Panthers 23-20 (59.4%)

Pick: Steelers +3.5 (50.5%)

Total: Over 41.5

Chicago at NY Jets

Prediction: Bears 21-20

Pick: Bears +2.5 (60.1%)

Total: Under 45.5

March Madness Projections Updated – March 5, 2013

Full Rankings

Projected Seeds

Number 1 Seeds: Gonzaga, Indiana, Michigan, Duke

Last 4 in: Boise State, Wichita State, Virginia, Stanford

Last 4 out: California, La Salle, Arizona State, Baylor

Full Tournament Projection

Screen Shot 2013-03-05 at 3.47.29 PMCheers.


From Deadspin: Nate Silver’s Braying Idiot Detractors Show That Being Ignorant About Politics Is Like Being Ignorant About Sports

This article by David Roher is fantastic: Nate Silver’s Braying Idiot Detractors Show That Being Ignorant About Politics Is Like Being Ignorant About Sports

The article also pointed me to the Princeton Election Consortium, which is also fantastic.  They have the probability of an Obama win at 99.0% and predict an electoral college win of 315-223. Below are some of the graphs they have produced about the election, I especially like the 2012 Electoral College Map with each state’s area displayed proportional to its electoral votes.


Tebow Mania and Passer Rating

Now that the college season is over, football fans can concentrate on what really matters: Tebow-mania!  Timmy Terrific has led the Denver Broncos to the AFC Divisional Round of the NFL playoffs.  The Broncos, who started the season 1-4, turned their season around by winning 7 of their last 11 games, many in dramatic fashion, with Mr. Tebow at the helm — good enough to squeak into the playoffs at 8-8 and even earn a home playoff game.  In that game, played Sunday, January 8th, they drew the heavily favored defending AFC champs the Pittsburgh Steelers, who took an early 6-0 lead.  Denver battled back with a big second quarter, but the Steelers made their own charge and ultimately the game went to overtime.  This was the first time a playoff game had gone to overtime since the inception of new NFL overtime rules.  Previously, overtime was sudden death, with the first team to score, either a touchdown or field goal (or safety), winning the game.  Under the new rules, only a touchdown on the first possession will end the game immediately; a field goal allows the other team a chance to possess the ball.  Needless to say, Denver won the coin toss (Pittsburgh called tails) and needed only one play to score a touchdown.  The play was an 80 yard pass over the middle that went the distance.One reason this occurred was because the Steelers had been bringing a lot of defenders close to the line of scrimmage, as they did not believe Tebow could beat them with his passing ability.  It was widely believed among the “experts” that Tebow, who is one of the greatest college football players of all time, and his style of play would not translate to success in the NFL.  Many people still believe this.

I hadn’t really thought much about Tebow one way or the other until one of his stats caught my eye.  In Denver’s week 10 win over the Kansas City Chiefs, Tebow’s passer rating was 102.6 based on 2 completions in 8 attempts, good for 69 yards, 1 touchdown, and no interceptions.  For some context, 102.6 was good enough for 7th best rating among starting quarterbacks that week.  This seemed odd to me since 2 for 8, 69 yards, and 1 touchdown seems like a terrible game.

So this got me wondering: What exactly is passer rating?  This website describes the formula in detail along with some of its history, but the basics are as follows.

1. Compute completions divided by passing attempts, subtract 0.3, and multiply by 5.
2. Compute yards divided by passing attempts, subtract 3, and multiple by 0.25
3. Compute touchdowns divided by passing attempts and multiply by 20.
4. Compute interceptions divided by passing attempts, multiply this by 25, and subtract this from 2.375

If any of the results of the four parts is less than 0 or greater than 2.375, that component is rounded up or down to the respective bound.  Now, add the four, possibly rounded, components together, multiply by 100, and divide by 6.  This yields a maximum score of 158.3.  (I swear I didn’t just make all of that up; the NFL actually uses this.)

Now since I like football and I love R, I decided to do some graphical exploring with passer rating.  Since the only topic anyone wants to talk about in the NFL right now is Tim Tebow, I figured I had to look at him.  And who better to compare him to than his opponent next week, three time Super Bowl champion Patriots quarterback Tom Brady.  Using the data from their regular season games (Tebow started 11 games and came in at half time in week 5; Brady started all 16), I created these graphs for Tom Brady and Tim Tebow.  Each individual graph shows how quarterback rating would vary based on number of completions and total passing yards for a fixed number of passing attempts, touchdowns, and interceptions. The green dot in each plot represents where each quarterback actually fell that week in their game.

What stands out to me in looking at these graphs is Brady’s consistency.  The green Brady dot seems to be always in the right, upper half of the graph.  Week in and week out he puts up around 300 yards (with the occasional 517 yard game thrown in) and a completion percentage in the mid to high 60s.  In fact, Brady had a completion percentage of over 50% in every single game this season.

Tebow, on the other hand, is, to put it politely, all over the place.  In week 13, Tebow put up a nearly perfect passer rating of 149.3, which is almost 14 points higher than Brady’s best passer rating of the season.  On the other hand, Tebow had a lower passer rating than Brady’s worst passer rating, 75.4, in 5 out of the 12 games Tebow started.  So you could say that almost half of the time this season, Tebow was worse than Brady’s worst.

This all adds up to the fact that the Broncos should lose to the Patriots.  Based on the stats, Brady is too good and Tebow is too inconsistent to amount to a Denver victory.  Of course, while you may find all of this interesting, in the end none of these numbers or pretty pictures mean anything at all to Tim Tebow, who, as they say, only cares about one stat and that’s winning.

Rule for Variance Inflation Factors

A quote from here:
“Goldberger (1991) notes that while the number of pages in econometrics
texts devoted to the problem of multi-collinearity in multiple regression is
large the same books have little to say about sample size. Goldberger states:
“Perhaps that imbalance is attributable to the lack of an exotic polysyllabic
name for ‘small sample size.’ If so, we can remove that impediment by introducing the term micronumerosity” (Goldberger, 1991: 248–249).”



NCAAB Rankings – 1/2/2012

Rankings as of 2:31pm on 1/2/2012.  Sagarin ratings as of 1/1/2012.  AP rankings as of 1/2/2012.

Previous rankings are here.

Pittsburgh, Oklahoma, and Miss St. fall out of the top 25 and New Mexico, North Carolina, and Purdue are in the top 25 this week.  The Mountain West conference has 3 teams in my top 25.  That is the same number as the ACC, Pac-12 and SEC COMBINED.  I’ve even given the MWC there own color (purple).  I considered giving them red since the Pac-12 wasn’t using it at all, but I think purple is a more fitting color for the MWC (think purple mountains majesty).

Breakdown by conference: 2, 6, 9, 4, 0, 1, 3

ACC Big East Big Ten Big 12 Pac 12 SEC MWC Other

Team Rank Change Record AP Sagarin
Syracuse 1 ↑1 15-0 1 1
Baylor 2 ↑2 13-0 4 6
Indiana 3 ↓2 13-1 12 4
Michigan St. 4 ↑5 13-2 10 8
Duke 5 ↑7 12-1 5 5
Ohio St. 6 ↓3 13-2 6 3
UConn 7 ↑10 12-1 8 26
Michigan 8 ↑8 12-2 16 41
Missouri 9 ↑4 13-0 7 11
Illinois 10 ↓3 12-3 11 47
Northwestern 11 ↓3 11-3 51
Marquette 12 ↓1 12-2 20 19
Wisconsin 13 ↓3 12-3 18 15
Georgetown 14 ↑11 12-1 9 10
Seton Hall 15 ↓1 12-2 38 28
UNLV 16 ↑3 15-2 17 12
Kansas State 17 ↓2 11-1 23 18
Purdue 18 NR 12-3 33 27
Kentucky 19 ↑4 13-1 2 2
North Carolina 20 NR 13-2 3 7
Minnesota 21 ↓16 12-3 50
Louisville 22 ↓16 12-2 11 13
San Diego St.  23 ↓5 12-2 24 38
Kansas 24 10-3 14 9
New Mexico 25 NR 12-2 34 31

BCS: My offer still stands…….if you want to contact me you can send me a tweet @StatsInTheWild.


Multidimensional Scaling, Republican Presidential Candidates, and “a douchebag”

If you don’t want to read this whole thing, just check out the graph: Multidimensional Scaling: Republican Candidates – 8/16/2011

I was having a conversation with some friends today and someone mentioned that Rick Perry might have problems in the election because there were rumors he was gay.  So I went to google and typed in “Rick Perry is” and google kindly offered me the following auto-complete options: “gay”, “an idiot”, “a rino“, “evil”, “not a conservative”.  This got me thinking how this compared with the other candidates google auto-completes.  For instance, if you google “Mitt Romney is” you get suggestions like “a mormon” and ” an idiot” as well as three other suggestions.  I did this for all of the major candidates (sorry Thaddeus) and recorded the five google auto-complete suggestions.

Then I created a vector for each candidate based on the google auto-complete words.  Each candidate was an observation and each word was a variable.  The candidate would get a 5 if the word was first on their list, a 4 if it was second, and so on with a 0 if the word was not mentioned in their auto-complete.

I then used multidimensional scaling (the cmdscale function in R) to allow me to visually display the relative positions of the candidates to each other.  This all led to this graphic: Multidimensional Scaling: Republican Candidates – 8/16/2011.  The location of the circles is based on multidimensional scaling, the size of the circle is relative to their standings in a national poll taken from, and the top five google auto-completes are displayed in or near the appropriate circle.

Some thoughts:

  • Every single candidate has the term “an idiot” in either the first or second auto-complete term
  • 3 candidates were listed as “hot” (Palin. Bachmann, and Romney)
  • “stupid” was only used to describe women
  • Perry and Santorum (who has a much bigger google problem that anything I’ve listed here) had “gay” listed in their autocpmpletes and Pawlenty had “definitely not gay”
  • Bachman and Palins circles are nearly identical in size (11.7% ad 11.4%, respectively) and words (they share “an idiot”, “hot”, and “stupid”)
  • “a douchebag” appears in auto-completes for Santorum, Gingrich, and Pawlenty.  I imagine it will be hard to win with this word attached to your name. (John Kerry couldn’t do it.)
  • The only overwhelmingly positive google auto-complete was for Herman Cain whose fifth auto-complete option was “awesome”
It can’t be good for Perry that he is so close to Pawlenty and Santorum, but he does have a significant amount of support at this point.  I’ll be interested to see how these Google auto-completes changes over time and with the polls.
For information on how Google auto-complete works, click here.