NFL Playoff Predictions

Wild Card Round

Texans (55.95%) over Bills, 23-20

Patriots (56.47%) over Titans, 24-20

Saints (65.16%) over Vikings, 28-20

Seahawks (53.52%) over Eagles, 24-22

Divisional Round

Ravens (69.03%) over Texans, 29-20

Chiefs (65.5%) over Patriots, 28-21

Saints (60.61%) over Packers , 28-22

49ers (64.6%) over Seahawks, 28-21

Conference Championship Games

49ers (51.52%) over Saints, 25-24

Ravens (50.5%) over Chiefs, 26-25

Super Bowl

Ravens (52.19%) over 49ers, 24-22





For posterity: 100 rappers better than Kanye

I don’t think Kanye is a very good rapper.  Other people disagree with me.  So I said I would name 100 rappers better than Kanye.  Here is my list:

1. Jay-Z

2. 2-Pac

3. Common

4. Eminem

5. Biggie

6. Nas

7-16. All of the Wu Tang clan

17-19. Beastie Boys

20-21. Outkast

22. Will-I-Am

23. Aesop Rock

24. Drake

25. Kendrick Lamar

26. Snoop Dog (barely)

27. Blackalicious

28. Danny Brown

29. Chris Webby

30. Spose

31. Pigeon John

32. Prof

33. Watsky

34. Dr. Dre

35-36. Run the jewels

37. Mc Chris

38. Ice Cube

39. Rakim

40. Krs-One

41. 50 Cent

42. Most Def

43. Xzibit

44. Talin Kweli

45. Black Thought

46. Q Tip

47. DMX

48. Busta Rhymes

49. Lauren Hill

50. Missy Elliot

51. Lil Kim

52. Childish Gambino

53. Nate Dogg

54. Phife Dog

55. Ali Shared Muhammed

56. Jarobi White

57. Logic

58. Chance the Rapper

59. Ludacris

60. Kid Cudi

61. Wiz khalifa

62. Hopsin

63. Nelly

64. MF Doom

65. Future

66. Nicki Minaj

67. Immortal Technique

68. Obie Trice

69. Brother Ali

70. J Dilla

71. Del the Funkee Homosapien

72. Diabolic

73. Kno

74. Natti

75. Deacon the Villain

76. Bizarre

77. 21 savage

78. Pusha T

79. Kodak Black

80. Meek Mill

81. Lil Baby

82. Gucci Mane

83. Jay Rock

84. Saba

85. Nipsey Hustle

86. Grandmaster Caz

87. Melle Mel 8

88. Big daddy Kane

89. Prodigy

90. Havoc

91. Twista

92. Rick Ross

93. 2 Chainz

94. Cardio B

95. Queen Latifah

96. Eve

97. M.I.A.

98. Lizzo

99. Salt

100. Pepa

The Cowboys are going to lose their way all the way to a home playoff game

Here is a fun plot that I made showing number of wins vs probability (according to  The 7-6 Bears, who just beat the 6-7 Cowboys last night, are probably going to miss the playoffs.  The, let me repeat this, 6-7 Cowboys are probably going to HOST a home playoff game.  I love it.  What I’m really rooting for though is for Washington to win the NFC East, get a home playoff game, and get blown out by 80.  That’s what I want for Christmas.  Thought what I really want to see someday is a team get a home playoff game with a 3-13 record!  It’s theoretically possible!

Screen Shot 2019-12-06 at 9.46.28 PM.png

College Football Playoffs

Here.  This is what I think will happen.


Google Image Search Series

Note: The full series can be viewed here.

Alright.  I’m really excited about this new series, which I’m calling the “Google Image Search Series”.  Below you’ll see some examples from this series and a brief explanation of how these are created.

How are these images made?

First, I choose a word or phrase such as “celebrity” or “famous person”, and I do an Google image search for that word.  I then take screen shots of the top 20 or so results.  This set of images becomes my training data.

The images are then converted from an image array to a vector with the first third of this vector representing the intensities of the red channel, the second third corresponding to the green channel, and the last third the blue channel.  So if the image was, for instance, 400 pixels x 400 pixels, the vector would be of length 400*400*3 = 480000 (because there are three color channels).

I then build one model for each element of this vector.  Every element in the vector is the response variables once, and everything else acts as predictor variables in a model.  Because of the n <<< p nature of this penalized regression models such as LASSO, Ridge, and elastic net were my first choice.  Once I have all of the models, I then use them to predict the red, green, and blue intensities at each pixel with initialized values of all intensities being 0.  Essentially this means starting with a solid black square and then letting the models modify each pixel accordingly.  Some of the results can be seen below:






I think the images above are amazing.  But others weren’t that interesting.  A lot of these ended up just looking like blurry messes. So I kept experimenting with different models.

What I realized is that I wanted a model that would do a BAD job recreating these images, because that would look more interesting.  So I turned to CART models.   CART models aren’t necessarily bad, but they certainly will not be complex enough to accurately capture all of the complexities of an image.  Also, as an additional way to further obscure the images, rather than working with the raw data as predictors, I’m using only the principle components as predictors.  This produces some really stunning images:

“Eye” (2019)


“Clown” (2019)


“Celebrity” (2019)


“Baseball” (2019)


“General” (2019)


“Cloud” (2019)


And here is the exciting part! I’ve just recently added these to my store and they can be purchased here and here (I needed to create two separate listings because I ran out of variants in the first).  Let me know if there are words or phrases you want to see.  Just send me a message on twitter @statsinthewild.


We are witnessing more really big spreads (15+ points) than usual in the NFL this season

So far this year, through 8 of 17 weeks (121 of 267 games) (i.e. less than half of the season), there have been EIGHT games in the NFL where the spread was 15 points or greater (and TWO(!!!) with spreads greater than 20):

  • Vikings (16.5) vs Washington Football Team (October 24, 2019)
  • Bills (-17) vs Dolphins (October 20, 2019)
  • Patriots (-16.5) vs Giants (October 10, 2019)
  • Patriots (-16.5) AT Washington Football Team (October 6, 2019)
  • Chargers (-15) AT Dolphins (September 29, 2019)
  • Cowboys (-21.5) vs Dolphins (September 22, 2019)
  • Patriots (-21) vs Jets (September 22, 2019)
  • Patriots (-18) AT Dolphins (September 15, 2019)

This seems like a lot!  So I went and checked back through 2006.  Turns out, this is a lot.  The most there has ever been in a FULL season is 14 in 2007.  Nine of those games were the Patriots during their undefeated season.  The next most in a season was 2009 with seven such games (mostly involving the Raiders…).  In 8 of the last 14 seasons, there were at most 3 such games (3 in 2006, 2011, 2016, 2017, 2018, 1 in 2014 and 2015, and ZERO(!) in 2010).

So we are ALREADY at 8 games and we are LESS THAN HALFWAY (there are NINE weeks left!) through the season.

Through 121 games 8 of them (or 6.61%) have had spreads of 15 or more.  The next two highest rates occurred in 2007 and 2009 at 3.75% and 2.62%, respectively.  No other season since 2006 has had more than 2% of games with spreads that large.  The plots below show survival curves of the spreads with all past seasons in blue and 2019 in red with the plot on the right zoomed in.  (The way to read this plot is that the height of the curve tell you the proportion of games played where the spread was LARGER than the x-value.)

Screen Shot 2019-10-29 at 2.06.09 PMScreen Shot 2019-10-29 at 2.07.23 PM

You can also see the increase in the number of high spread games in this plot below, which shows the density estimates of the absolute value of the spreads.

Screen Shot 2019-10-29 at 2.04.21 PM

Notice that the proportion of games with spreads around 10 are actually lower than most years, but as you get out past 15, you see a big jump in the density.  If instead we look at the actual spread rather than the absolute value, we get the plot below.

Screen Shot 2019-10-29 at 2.03.14 PM

Here, negative values correspond to a home team being favored and positive values indicate an away favorite.  2019 is shown in black.  Notice the difference in the behavior in the tails of the distribution in 2019.  There are both more big home favorites but also more big AWAY favorites.  The big away favorites are particularly interesting.  Since 2006 there have only been 8 total games where the away team was favored by 15 or more points and SIX of those games involved the Patriots:

  • Patriots (-16.5) at Washington Football Team (October 6, 2019)
  • Chargers (-15) at Dolphins (September 29, 2019)
  • Patriots (-18) at Dolphins (September 15, 2019)
  • 49ers (-16) at Jaguars (September 15, 2013)
  • Patriots (-15.5) at Buccaneers (October 25, 2009)
  • Patriots (-19) at Ravens (December 3, 2007)
  • Patriots (-16.5) at Bills (November 18, 2007)
  • Patriots (-16) at Dolphins (October 21, 2007)

So, in summary, between 2006 and 2018 we witnessed 5 total games where the away team was favored by 15 or more.  That’s an average on ONE game every 2.6 seasons, but this season we are seeing an away team favored by 15 or more once every 2.67 WEEKS!

We’ll see if this trend continues for the rest of the 2019 season, but so far we are seeing a lot more extremely lopsided games than usual.




NFL Picks – Week 7

So I’m way behind on updating these.  I blame my kids.  Go Bears.

You can view my predictions here.  Go Bears.

Season: SU: 38-24-1, Spread: 34-29, O/U: 31-29-3

Week 1: SU: 10-5-1, Spread: 9-7, O/U: 8-8

Week 2: SU: 9-7, Spread: 9-7, O/U: 6-10

Week 3: SU: 11-5, Spread: 9-7, O/U: 11-2-3

Week 4: SU: 8-6, Spread: 6-8, O/U: 6-8

Week 5: SU: 1-0, Spread: 0-1, O/U: 1-0

Prediction: Chiefs 28-25

Prediction: Rams 27-26

Prediction: Bills 29-12

Prediction: Jaguars 21-20

Prediction: Texans 25-22

Prediction: Vikings 23-21

Prediction: Packers 27-17

Prediction: Giants 25-23

Prediction: 49ers 27-19

Prediction: Titans 21-20

Prediction: Bears 23-17

Prediction: Seahawks 22-21

Prediction: Cowboys 24-23

Prediction: Patriots 27-12

NFL Picks – Week 6

You can view my predictions here.  Go Bears.

Season: SU: 38-24-1, Spread: 34-29, O/U: 31-29-3

Week 1: SU: 10-5-1, Spread: 9-7, O/U: 8-8

Week 2: SU: 9-7, Spread: 9-7, O/U: 6-10

Week 3: SU: 11-5, Spread: 9-7, O/U: 11-2-3

Week 4: SU: 8-6, Spread: 6-8, O/U: 6-8

Week 5: SU: 1-0, Spread: 0-1, O/U: 1-0

Prediction: Patriots 29-20
Pick: Giants +17
Over/Under: Over 43

Prediction: Panthers 25-24

Prediction: Browns 22-21

Prediction: Jaguars 25-22

Prediction: Chiefs 31-24

Prediction: Dolphins 23-22

Prediction: Vikings 25-24

Prediction: Ravens 26-21

Prediction: Falcons 26-21

Prediction: Rams 32-17

Prediction: Broncos 22-21

Prediction: Cowboys 21-20

Prediction: Chargers 27-26

Prediction: Packers 27-23

NFL Picks – Week 5

These are a week late.  But before you accuse me of cheating, see how I did. #notgood

You can view my predictions here.  Go Bears.

Season: SU: 44-31-1, Spread: 40-37, O/U: 37-36-3

Week 1: SU: 10-5-1, Spread: 9-7, O/U: 8-8

Week 2: SU: 9-7, Spread: 9-7, O/U: 6-10

Week 3: SU: 11-5, Spread: 9-7, O/U: 11-2-3

Week 4: SU: 8-6, Spread: 6-8, O/U: 6-8

Week 5: SU: 6-9, Spread: 6-8, O/U: 6-7

Prediction: Rams 28-20
Pick: Rams +1.5
Over/Under: ????

Prediction: Panthers 21-19
Pick: Jaguars +3.5
Over/Under: Under 40

Prediction: Bengals 26-17
Pick: Bengals -3
Over/Under:Under 47.5

Prediction: Falcons 26-25
Pick: Falcons +4
Over/Under: Over 50.5

Prediction: Saints 30-26
Pick: Saints -3
Over/Under:Over 46

Prediction: Vikings 24-22
Pick: Giants +5
Over/Under: Over 43.5

Prediction: Titans 22-16

Prediction: Eagles 25-18
Pick: Jets +14
Over/Under: Under 43.5

Prediction: Steelers 27-24
Pick: Steelers +3.5
Over/Under: Under 44

Prediction: Bears 22-20
Pick: Raiders +5
Over/Under: Over 40

Prediction: Patriots 27-22
Pick: Washington Football Team +15.5
Over/Under:Over 42

Prediction: Chargers 29-20
Pick: Chargers -6
Over/Under: Over 44.5

Prediction: Cowboys 23-21
Pick: Packers +3.5
Over/Under: Under 47

Prediction: Chiefs 33-22
Pick: Colts +11
Over/Under: Under 56

Prediction: 49ers 22-21
Pick: Browns +4
Over/Under: Under 46.5

Undefeated NFL Matchups with large spreads

The 3-0 New England Patriots will play the 3-0 Buffalo Bills this week.  However, in spite of the fact that both teams are undefeated, the Patriots are favored by 7.5 points.  So this got me wondering: Has there ever been a biggest spread in a game that featured two undefeated teams?  So I went back and checked.  (Well, I went back through 2006 because that’s how far back I have spread data readily available….)

First of all, I’m not counting games in week 1 where all the teams are “undefeated”.  And I’m not counting matchups where both teams are 1-0.  Everything else counts.

So, for a 2-0 vs 2-0 I found a Steelers-49ers game from 2007-09-23 where Pittsburgh was favored by 10 points (Steelers won 37-16).  For 3-0 vs 3-0, I found a game from October 4, 2009 between the Saints and the Jets where the Saints were favored by 7.5.  On September 30, 2013 there was another 3-0 vs 3-0 game the a 7 point spread: The Saints (again) were favored by 7 over the Miami Dolphins.

Of note, on Sep 25, 2011 the 2-0 Patriots were favored by7 points over the 2-0 Buffalo Bills.  (The Bills went on to win 34-31).

Other 2-0 vs 2-0 match-ups with big spreads include:

Colts favored by 6.5 over the Texans on Sunday Sep 23, 2007

Colts favored by 6.5 over the Jaguars on Sunday Sep 24, 2006

Finally, I have to mention that on Sunday Nov 4, 2007, the 8-0 Patriots played the 7-0 Colts and the Patriots were favored by 5.5 points.





Loading cart ...