SITE NEWS: We are moving all of our site and company news into a single blog for Sports-Reference.com. We'll tag all PFR content, so you can quickly and easily find the content you want.

Also, our existing PFR blog rss feed will be redirected to the new site's feed.

Pro-Football-Reference.com ยป Sports Reference

For more from Chase and Jason, check out their work at Football Perspective and The Big Lead.

The Rivers Index, Playoffs Edition (2001-2010)

Posted by Neil Paine on January 21, 2011

Back in November, I developed what I called "The Rivers Index" (so named for Philip Rivers), a metric that measured how many games a QB should have won based on nothing more than his own passing performance. Today, I'm going to apply that same concept to the last 10 years of playoff competition, this time using 10 years of data and adjusting for opposing defenses + weather.

Here's how it worked: For every year from 2001-2010, I gathered team regular-season passing gamelog data and playoff QB gamelog data. I then ran an SRS-style process on each regular season dataset to produce offensive and defensive team Adjusted Yards per Passing Attempt ratings, plus homefield effects on AY/A. I also added Doug's weather effects into the mix, and calculated the AY/A a league-average QB would be expected put up in any given game. I called the difference between that and the team's actual offensive AY/A their "AY/A above average" (AYAAA).

Using all 5100 games from the 2001-2010 period, I ran a logistic regression between team wins (win=1, loss=0) and their AYAAA in the game. The result is an equation that predicts how often a team "should" win, given it's QB play relative to the league average (adjusted for location, opponent, & weather):

xWin% ~ =1/(1+EXP(-0.3424889*AYAAA))

I then isolated every QB who was their team's leading pass-attempter in a playoff game since 2001, and plugged their numbers into the equation to determine their expected winning % in each game.

Let's take Jay Cutler vs. Seattle last weekend as an example: Cutler was 15-28 for 274 yards, 2 TD, and 0 INT, which works out to an AY/A of 11.21. The 2010 NFL average was 6.54 AY/A, the homefield effect for both the offense & defense combined was 0.31, Seattle's pass defense was 1.29 AY/A worse than average, and the game was played in "moderately cold" Chicago in January, yielding a weather adjustment of -0.3. Here's the league-average expectation in Cutler's game:

LgAvg = 6.54 + 0.5*0.31 +1.29 - 0.3 = 7.69

Cutler's 11.21 was 3.52 better than the league-average expectation, so we plug 3.52 into the logistic equation and estimate that a QB with Cutler's performance should win 77% of the time.

After doing this for every QB who was his team's leading passer in a playoff game from 2001-2010, here are the results, sorted by games:

(Note: only results from team-leading games are included.)

Player Gms Won Lost WPct xWins xLosses xWin% PPG Cmp Att Yds TD Int AY/A LgAvg AYAAA
Tom Brady 18 13 5 0.722 10.42 7.58 0.579 23.9 412 664 4292 30 16 6.28 5.50 0.78
Peyton Manning 17 9 8 0.529 10.39 6.61 0.611 23.1 417 644 4968 28 19 7.26 5.50 1.76
Donovan McNabb 14 8 6 0.571 7.77 6.23 0.555 21.7 297 503 3410 21 15 6.27 5.65 0.62
Matt Hasselbeck 11 5 6 0.455 6.30 4.70 0.573 25.1 237 406 2741 18 9 6.64 5.86 0.78
Ben Roethlisberger 11 9 2 0.818 6.54 4.46 0.595 27.2 191 310 2465 17 12 7.31 5.81 1.50
Brett Favre 10 4 6 0.400 5.45 4.55 0.545 24.0 211 342 2465 19 18 5.95 5.86 0.09
Kurt Warner 9 6 3 0.667 6.26 2.74 0.696 30.4 206 301 2524 20 7 8.67 5.67 3.00
Jake Delhomme 8 5 3 0.625 5.20 2.80 0.650 22.5 130 226 1847 12 10 7.24 5.42 1.83
Drew Brees 7 4 3 0.571 4.31 2.69 0.615 28.7 189 285 2052 15 2 7.94 6.61 1.33
Joe Flacco 7 4 3 0.571 2.95 4.05 0.422 20.6 98 184 1050 4 7 4.43 5.22 -0.79
Eli Manning 7 4 3 0.571 3.73 3.27 0.533 16.6 113 193 1297 8 7 5.92 5.50 0.42
Philip Rivers 7 3 4 0.429 4.58 2.42 0.654 19.9 134 229 1820 8 9 6.88 4.76 2.12
Jeff Garcia 6 2 4 0.333 3.06 2.94 0.509 20.2 126 217 1357 7 7 5.45 5.47 -0.02
Chad Pennington 6 2 4 0.333 3.35 2.65 0.558 18.8 132 216 1418 8 8 5.64 5.17 0.47
Rich Gannon 5 3 2 0.600 3.29 1.71 0.657 28.6 113 175 1294 10 6 6.99 5.05 1.94
Steve McNair 5 2 3 0.400 2.57 2.43 0.514 19.6 98 158 1074 5 8 5.15 4.82 0.33
Mark Sanchez 5 4 1 0.800 3.09 1.91 0.618 20.6 75 124 922 7 3 7.48 5.91 1.57
Michael Vick 5 2 3 0.400 2.60 2.40 0.521 21.2 78 139 901 4 4 5.76 5.26 0.50
Rex Grossman 4 2 2 0.500 1.68 2.32 0.421 26.0 69 133 783 4 4 5.14 6.06 -0.92
Brad Johnson 4 3 1 0.750 1.95 2.05 0.488 28.8 75 134 872 5 7 4.90 5.22 -0.32
Jake Plummer 4 1 3 0.250 1.74 2.26 0.436 19.5 80 120 885 5 6 5.96 6.66 -0.70
Tony Romo 4 1 3 0.250 1.79 2.21 0.448 18.5 80 135 832 4 2 6.09 6.81 -0.72
Marc Bulger 3 1 2 0.333 1.77 1.23 0.589 22.3 68 113 944 4 5 7.07 6.26 0.81
Aaron Rodgers 3 2 1 0.667 2.47 0.53 0.822 38.0 77 105 969 10 1 10.70 6.02 4.68
Mark Brunell 2 1 1 0.500 0.70 1.30 0.349 13.5 29 52 283 1 1 4.96 5.99 -1.03
Kerry Collins 2 0 2 0.000 1.30 0.70 0.649 24.0 55 85 623 4 2 7.21 5.34 1.87
Daunte Culpepper 2 1 1 0.500 1.38 0.62 0.691 22.5 43 75 600 5 2 8.13 5.91 2.22
David Garrard 2 1 1 0.500 1.09 0.91 0.543 25.5 31 54 418 3 3 6.35 5.17 1.18
Elvis Grbac 2 1 1 0.500 0.89 1.11 0.444 15.0 30 55 286 1 3 3.11 5.60 -2.49
Trent Green 2 0 2 0.000 0.85 1.15 0.424 19.5 32 54 319 2 2 4.98 5.80 -0.82
Tommy Maddox 2 1 1 0.500 1.15 0.85 0.573 33.5 51 89 633 5 3 6.72 5.82 0.90
Matt Ryan 2 0 2 0.000 0.55 1.45 0.277 22.5 46 69 385 3 4 3.84 6.91 -3.07
Kordell Stewart 2 1 1 0.500 0.85 1.15 0.424 22.0 36 64 409 1 4 3.89 5.39 -1.50
Drew Bledsoe 1 1 0 1.000 0.51 0.49 0.509 24.0 10 21 102 1 0 5.81 5.71 0.10
Quincy Carter 1 0 1 0.000 0.35 0.65 0.348 10.0 21 36 154 0 1 3.03 4.87 -1.84
Matt Cassel 1 0 1 0.000 0.05 0.95 0.049 7.0 9 18 70 0 3 -3.61 5.03 -8.64
Todd Collins 1 0 1 0.000 0.35 0.65 0.347 14.0 29 50 266 2 2 4.32 6.16 -1.84
Jay Cutler 1 1 0 1.000 0.77 0.23 0.770 35.0 15 28 274 2 0 11.21 7.69 3.52
Jay Fiedler 1 0 1 0.000 0.28 0.72 0.285 3.0 15 28 122 0 1 2.75 5.44 -2.69
Kelly Holcomb 1 0 1 0.000 0.86 0.14 0.855 33.0 26 43 429 3 1 10.33 5.14 5.19
Tarvaris Jackson 1 0 1 0.000 0.26 0.74 0.259 14.0 15 35 164 0 1 3.40 6.47 -3.07
Jon Kitna 1 0 1 0.000 0.32 0.68 0.318 17.0 24 40 197 1 2 3.18 5.40 -2.22
Byron Leftwich 1 0 1 0.000 0.31 0.69 0.313 3.0 18 31 179 0 1 4.32 6.61 -2.29
Shane Matthews 1 0 1 0.000 0.09 0.91 0.095 19.0 8 17 66 0 2 -1.41 5.18 -6.59
Carson Palmer 1 0 1 0.000 0.47 0.53 0.470 14.0 18 36 146 1 1 3.36 3.71 -0.35
Chris Simms 1 0 1 0.000 0.27 0.73 0.272 10.0 25 38 198 0 2 2.84 5.72 -2.88
Vinny Testaverde 1 0 1 0.000 0.74 0.26 0.741 24.0 26 40 268 3 0 8.20 5.13 3.07
Anthony Wright 1 0 1 0.000 0.33 0.67 0.330 17.0 20 37 214 1 2 3.89 5.96 -2.06
Vince Young 1 0 1 0.000 0.39 0.61 0.390 6.0 16 29 138 0 1 3.21 4.51 -1.30

The individual games file can be found here.

The Rivers Index is about the biggest differences between deserved W-L and actual W-L, so here are the QBs again, sorted from biggest overperformers to biggest underperformers:

Player Gms Won xWins Diff
Tom Brady 18 13 10.42 2.58
Ben Roethlisberger 11 9 6.54 2.46
Brad Johnson 4 3 1.95 1.05
Joe Flacco 7 4 2.95 1.05
Mark Sanchez 5 4 3.09 0.91
Drew Bledsoe 1 1 0.51 0.49
Rex Grossman 4 2 1.68 0.32
Mark Brunell 2 1 0.70 0.30
Eli Manning 7 4 3.73 0.27
Jay Cutler 1 1 0.77 0.23
Donovan McNabb 14 8 7.77 0.23
Kordell Stewart 2 1 0.85 0.15
Elvis Grbac 2 1 0.89 0.11
Matt Cassel 1 0 0.05 -0.05
David Garrard 2 1 1.09 -0.09
Shane Matthews 1 0 0.09 -0.09
Tommy Maddox 2 1 1.15 -0.15
Jake Delhomme 8 5 5.20 -0.20
Tarvaris Jackson 1 0 0.26 -0.26
Kurt Warner 9 6 6.26 -0.26
Chris Simms 1 0 0.27 -0.27
Jay Fiedler 1 0 0.28 -0.28
Rich Gannon 5 3 3.29 -0.29
Drew Brees 7 4 4.31 -0.31
Byron Leftwich 1 0 0.31 -0.31
Jon Kitna 1 0 0.32 -0.32
Anthony Wright 1 0 0.33 -0.33
Todd Collins 1 0 0.35 -0.35
Quincy Carter 1 0 0.35 -0.35
Daunte Culpepper 2 1 1.38 -0.38
Vince Young 1 0 0.39 -0.39
Aaron Rodgers 3 2 2.47 -0.47
Carson Palmer 1 0 0.47 -0.47
Matt Ryan 2 0 0.55 -0.55
Steve McNair 5 2 2.57 -0.57
Michael Vick 5 2 2.60 -0.60
Vinny Testaverde 1 0 0.74 -0.74
Jake Plummer 4 1 1.74 -0.74
Marc Bulger 3 1 1.77 -0.77
Tony Romo 4 1 1.79 -0.79
Trent Green 2 0 0.85 -0.85
Kelly Holcomb 1 0 0.86 -0.86
Jeff Garcia 6 2 3.06 -1.06
Kerry Collins 2 0 1.30 -1.30
Matt Hasselbeck 11 5 6.30 -1.30
Chad Pennington 6 2 3.35 -1.35
Peyton Manning 17 9 10.39 -1.39
Brett Favre 10 4 5.45 -1.45
Philip Rivers 7 3 4.58 -1.58

Unbelievably enough, Philip Rivers is actually dead last here as well, merely reinforcing his status as the metric's namesake. Much like his regular-season frustrations, he was a .429 playoff record despite playing like a .654 quarterback.

At the other end, Tom Brady's recent losses still haven't caused his deserved performance to catch up to his actual W-L. He was obviously due for a correction after winning his first 10 playoff games as a starter (this system considers it 9 straight, since Drew Bledsoe threw more passes in the 2001 AFC title game); during those 9 games, he played like a .625 QB. Since the 2004 Super Bowl, Brady has a .444 record and has played like a .532 QB, so his luck has actually swung the other way in recent seasons.

Meanwhile Ben Roethlisberger, #2 on the list, has a chance to pad his Rivers Index even more this weekend. His opponent, Mark Sanchez, is a budding Rivers Index phenom in his own right. Sanchez sports an .800 record as his team's leading passer in the playoffs, despite playing like a .618 QB. Ironically, in that sense his career arc could very well follow that of a young Brady.

This entry was posted on Friday, January 21st, 2011 at 12:05 pm and is filed under Quarterbacks, Simple Rating System, Statgeekery. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.