On Doug Pederson, The Front Office and The Quarterback, Part 3

The Chip Kelly Effect

While we cannot draw any definitive, concrete conclusions here, we would be hard pressed to ignore that all four quarterbacks that started for Chip Kelly saw an improvement — sometimes considerably — in their production compared to their career averages under other head coaches.

Here are Bradford’s 2015 totals compared with his career numbers. The difference is highlighted at the bottom.

Sam Bradford Cmp% Y/G TD/G INT/G TD% INT% Y/A NY/A QB Rating
Kelly 65.0 266.07 1.35 1 3.6 2.6 7.0 6.53 86.4
Other 58.6 225.81 1.20 .77 3.4 2.2 6.29 5.99 79.3
Difference 6.4+ 40.26+ .15 .23+ .2+ .4+ .71+ .85+ 7.1+

The only area of regression is interceptions per game, where Bradford saw a .4% increase in his interception rate. But otherwise, Bradford had arguably the best season of his career in Philadelphia.

The same can be said of Mark Sanchez, Nick Foles, and Michael Vick:

Mark Sanchez Cmp% Y/G TD/G INT/G TD% INT% Y/A NY/A QB Rating
Kelly 64.3 233.38 1.38 1.15 4.5 3.8 7.59 6.55 86.6
Other 55.1 195.03 1.09 1.11 3.6 3.7 6.48 5.66 71.7
Difference 9.2+ 38.35+ .29+ .04+ .9+ .1+ 1.11+ 1.26+ 14.91+
Nick Foles Cmp% Y/G TD/G INT/G TD% INT% Y/A NY/A QB Rating
Kelly 61.9 240.66 1.9 .57 6.4 1.9 8.05 7.23 100.5
Other 58.3 208.38 .72 .83 2.2 2.5 6.23 5.54 73.4
Difference 3.6+ 32.28+ 1.18 -.25 4.2+ -.6 1.82+ 1.69 27.05+
Michael Vick Cmp% Y/G TD/G INT/G TD% INT% Y/A NY/A QB Rating
Kelly 54.6 173.85 .71 .42 3.5 2.1 8.62 7.15 86.5
Other 56.2 156.24 .94 .62 4.2 2.8 6.91 5.77 80.1
Difference -1.6 17.61+ -.23 -.2 -.6 -.6 1.71+ 1.39+ 6.4+

Take a look at these charts, courtesy of David Menard, which show how each quarterback improved over their career numbers in key stats. Bradford is in blue, Sanchez orange, Vick gray, and Foles yellow:

Bradford 1

The baseline 0.0% is the career averages of each quarterback. So, we see that Bradford saw a 6.4% increase in his completion percentage playing under Kelly, Nick Foles saw a 27% increase in his quarterback rating, and so on and so forth.

Here is another graph which highlights some more advanced metrics from Pro-Football-Reference.com: Y/A, NY/A, AY/A, and ANY/A:

Bradford 2

On average, a quarterback playing under Kelly saw an increase of: 4.4% in completion percentage, 32.12 passing yards per game, .35 touchdowns per game (or 5.56 total touchdowns over 16 games), 1.33 YPA, and 13.86 points in QB Rating, along with a reduction of interceptions by .7%. 

Now to be clear: we cannot say that these increases across the board are solely attributable to Chip Kelly. There are a number of factors that play into this improvement, including the quality of teammates and opponents, to name a few.

But, we cannot dismiss these numbers as sheer happenstance either. For all the talk about Kelly’s offense being figured out by the NFL, it was considered “quarterback friendly” for a reason: from the simplified verbiage, to the predefined reads, to the mismatches created by tempo (when it was working), Kelly put his quarterbacks in position to make easy plays. And those quarterbacks produced well above their career averages as a result.

Which makes it hard to determine just how much of Bradford’s improvement was a result of Sam Bradford, compared to how much of it was attributable to playing in Kelly’s system.

Approximate Value

Unless you break down the tape of most NFL quarterbacks, it is easy to watch Bradford’s play and convince yourself he is a top 10 quarterback. Every so often, Bradford flashes the tools that have excited scouts and general managers for years, making you convinced that Bradford is the real deal.

Which is why advanced metrics can be handy. They have a way of cutting through our confirmation bias and telling us how a player truly stacks up against the competition. Enter Pro-Football-Reference.com’s Approximate Value metric.

Approximate Value is similar to the Win Shares stat derived by the venerable Bill James in baseball. That is, Approximate Value attempts to attach a single number — or value/ranking, if you will — to every player that has played football since 1950. The idea is to determine how much value each player has added to his team’s ability to win games relative to his peers. It is by no means a perfect statistic, and PFR cautions against using this as a pure “ranking.” But it still helps contextualize how a player compares to other quarterbacks over an extended period of time.

I compiled the following search to put Bradford’s career to date in perspective (click the hyperlink to see the entire list): For combined seasons, from 1990 to 2015, from 1st to 6th season, played QB, sorted by descending Approximate Value.

What does that mean, exactly? This list ranks the first six seasons of every quarterback — in terms of the value that they added to their team —  that has played over the last 25 years. Of the 100 qualifying quarterbacks, Bradford ranks almost dead center: 55th overall.

Here is his company:

  • 49. Jon Kitna
  • 50. Scott Mitchell
  • 51. Brad Johnson
  • 52. Mark Rypien
  • 53. Matt Cassell
  • 54. Matt Schaub
  • 55. Sam Bradford
  • 56. Kyle Orton
  • 57. Mark Sanchez
  • 58. Vince Young
  • 59. Jay Fielder
  • 60. Byron Leftwhich


The knee jerk reaction here might be to dismiss these numbers as an unfair assessment of Bradford’s career given the bad situation he found himself in while at St. Louis: no skill position players, bad offensive line, revolving door of offensive coordinators, etc. But you can come up with many of the same excuses for each of these quarterbacks: Sanchez was stuck on Rex Ryan’s offensiveless team for years; Vince Young had zero help in Tennessee; Byron Leftwhich spent the majority of his career on the anemic Jacksonville Jaguars.


But more importantly, these numbers, to an extent, isolate the static and provide the actual contribution that a quarterback provides to his team, regardless of his surrounding environment.

If we break the advanced metrics upon which PFR relies, we can see why Bradford has such a bad rating. For these advanced numbers below, 100 is considered average production by an NFL starter.

Year Y/A+ NY/A+ AY/A+ ANY/A+ Cmp%+ TD%+ INT%+ Rate+
2010 81 84 87 88 98 86 106 92
2011 79 73 86 82 75 71 119 82
2012 92 94 96 97 96 95 107 97
2013 83 89 100 102 96 111 119 105
2014 94 98 92 95 109 88 97 95

In other words, from an advanced metric standpoint, Bradford is and has been producing at a below average rate almost across the board for his entire career.

But without context, its hard to appreciate what these numbers represent. So I rank another search: From 1990 until 2015, among all of the quarterbacks that have attempted at least 1,000 passes. There are 119 quarterbacks that qualify, here are Bradford’s ranks:

Y/A+ NY/A+ AY/A+ ANY/A+ Cmp%+ TD%+ INT%+ Rate%+
110/119 108/119 94/119 98/119 72/119 111/119 23/119 88/119

I know these numbers are getting a little esoteric at this point; but the big picture take-away is that Bradford is contributing less value on average from each throw that he makes than the significant majority of quarterbacks that have played in the last 25 years.

Refusal To Take Shots Deep

Bradford’s refusal to take shots deep is one of the primary reasons behind his poor ranking. And while playing risk adverse football improves your completion percentage, it hurts your teams ability to score points.

Now, Bradford’s supporters are likely about to jump down my throat, since Bradford completed the 4th most passes of 40+ yards this season.

But that statistic is wholly misleading. Deep passes are defined as any pass that travels more than 20+ yards. And when we scale it back to its original definition, Bradford comes up considerably short. Only 10.2 percent of his pass plays were deep throws, which was 29th of 35 qualifying quarterbacks, according to PFF.com.

If you were wondering why the Eagles failed to put up big offensive numbers when Bradford’s play improved, this is the primary reason. In order to score points, offenses must take risks. The numbers bear out that the more risks you take — especially throwing deep — the more points you can expect to score. As Advanced Football Analytics laid out a few years back, a short pass has an average expected points added (EPA) of -0.01, with a success rate of approximately 47%. But deep passes have an EPA of +.45, with a success rate of 43%.

Bradford’s risk adverse approach limited the Eagles opportunities to score points. And it is that approach that has also contributed to his below average rankings across the board from a value perspective.


Perhaps Sam Bradford really is a victim of circumstance and we saw him starting to tap into his true potential down the stretch of the 2015 season. He wouldn’t be the first quarterback to follow that career arc. Alex Smith, Brad Johnson, and Mark Brunell all started their careers off slowly only to turn things around.

But there is simply too much data that suggests Bradford is a below average quarterback for me to feel comfortable locking him up long term. That is especially true when you consider that Bradford has played in only 63 of 96 possible games in his career.

So the best bet is for the Eagles to franchise tag Bradford. Doing so gives them a number of options to solve the quarterback position, from further evaluating Bradford for the 2016 season, to trading him for a valuable draft pick. Franchising Bradford also does not prevent the Eagles from drafting a quarterback this year. In fact, they must draft at least one quarterback regardless of what they do with Bradford. The important thing is that the Eagles maintain flexibility until they are sure they have the solution in house. Franchising Bradford is the best way to accomplish that goal.

4 thoughts on “On Doug Pederson, The Front Office and The Quarterback, Part 3

  1. Excellent use of statistics to support your analysis. I have to clarify one thing: trading Bradford is not an option as long as he is a free agent. They can sign him or tag him and he is tradeable. If he is not signed by the Eagles by the respective deadline(s), he is free to sign elsewhere and the Eagles get zilch in return.

    • Glad you liked the article. My comment about trading Bradford was premised on the notion that they would first franchise him. Obviously, cannot trade a free agent!

  2. Onsider the possibility that the reason he didn’t throw deep well enough has something to do with a slow receiving corps that couldn’t get open deep.

  3. Pingback: Five thoughts on the preseason | Eagles Rewind

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s