August 15, 2014

The Diff: Coveted Chip Kelly brings Moneyball style to NFL

The Diff is your weekly WFNY look into the amazing world of sports statistics. For a complete log of articles, click this link. Last week, I wrote about some Browns storylines heading into Week 1. Now, I’m writing about the NFL’s latest fad and one of my favorite topics ever: Chip Kelly.

The Diff

Jan. 5-6, 2013: A crazy weekend in Browns news. The next Monday, the WFNY crew congregated for a roundtable to reflect on what exactly just happened. In the end, University of Oregon football coach Chip Kelly had indeed bolted for the NFL, but chose the Philadelphia Eagles instead of Cleveland or elsewhere. I was pretty decimated; not only because I was on Twitter alert for 16 hours a day for naught, but because the much-coveted Chip Kelly had gone elsewhere. My favorite football coach wouldn’t be coaching my favorite team.

That weekend, I had one backup focus to all of the Twitter news and latest “reports” from the negotiating dinner tables in Arizona: Absorb everything possible about one Chip Kelly. A tweet, an article, an essay, a profile, whatever it was, I planned on reading it. I took in as much as I possibly could and quickly drafted over 1,000 words on a piece titled “Myths and Facts: Tying up the loose ends on Chip Kelly”. It was also for naught, and I was forced instead to write that weekend about Brett Myers and the 2013 NBA Draft. Now, I’ll share some of those remnants.

That January week, the Browns were said to have interviewed eventual Bills coach Doug Marrone, eventual Bears coach Marc Trestman and many, many others. That Friday, they eventually shocked the region when they announced Rob Chudzinski as the new head coach. Yet one name remained on the minds of many Northeast Ohio fans as a lost opportunity in Cleveland: Chip Kelly.

Fast forward eight months, just after another woeful Browns loss, and the Eagles are playing their opening game on Monday Night Football against the reigning NFC East champs. Their electric first half takes over Twitter and the NFL community by storm: Philadelphia ran a shocking 53 plays in 30 minutes en route to a 26-7 lead. Eventually, they held on for the 33-27 victory over the Washington Redskins. They ran just 77 plays in the game, slowing down their frenetic pace in the second half.

Background of success

To know 49-year-old Chip Kelly’s current status as NFL innovator and the latest to shun Cleveland, it’s relevant to share his history as an elite coaching prospect. He started out as a high school quarterback, University of New Hampshire cornerback and high school offensive coordinator. Then, he jumped around various mostly defensive-oriented assistant coaching jobs in East Coast colleges, proving he isn’t just a offense-only coach.

Throughout all these years, he soaked up all of the information he could. He would attend clinics, information sessions and conferences to learn from anyone he could. He added bits and pieces of information into his coaching footprint from dozens and dozens of places. Then, in 2007, after multiple record-breaking seasons as the New Hampshire coach, he got his first “big” break to become Mike Bellotti’s offensive coordinator at Oregon. When Bellotti became the athletic director in 2009, Kelly was promoted to head coach.

The Eagles website Iggles Blitz had a great breakdown last December of Kelly’s track to being the one that other coaches sought out for advice. And these weren’t just ordinary coaches: This was Jon Gruden, Bill Belichick and Pete Carroll. Some NFL and collegiate champs, some of the best in the business. His up-tempo non-stop offense quickly gained a reputation, leading to a major Pete Thamel New York Times article in 2010.

But to say Kelly designed the first fast-paced offense isn’t exactly accurate. A 2008 Rivals article breaks down how an even more extreme Arkansas high school sought out a computerized system for various football odds. It resulted in a team-wide decision to avoid punting at all times. This was a team that was causing headaches for its opponents for years, albeit less popularized than Kelly’s success in Oregon, and before that, New Hampshire and elsewhere. More on this math-based approach to football later.

Running-based attack

NFL fans received their introduction to Chip Kelly’s offensive style in the Eagles’ shocking win on Monday night. The damage: a career-high 31 carries for LeSean McCoy for 184 yards. Overall, Philadelphia ran on 49 of those 77 plays, preferring a even more conservative style in the second half with the lead. DeSean Jackson also had an impressive game with seven catches for 104 yards and a touchdown.

The Oregon football analysis website Fish Duck had a very intricate multi-part tutorial series on the team’s spread offense last year. In these various photos and videos, based on reports from the team itself, one can see the multi-look running plays that were a focus of the Ducks’ multi-year success in college football. The Inside and Outside Zone Read were the basic initial building blocks.

Below, one can see a chart of Oregon’s offensive statistics during Kelly’s six years on staff. The 2007-2008 seasons were as offensive coordinator and the final four were as the head coach:

chip kelly oregon

One can note how in 79 total games, Oregon ran 75.6 plays from scrimmage per game. Of those plays, nearly 62% were running plays. No NFL team has exceeded a rate of 54% running plays over the past three seasons. The Ducks also averaged 5.94 yards per these carries, which was right on par with what the Eagles accomplished on Monday night. These teams were built on running the football — not exceptionally so per se with the quarterback, despite public opinion — and accurate passing.

In September 2012, The Oregonian’s Adam Jude wrote on the team’s breakneck speed and the records it was setting. Once, Kelly even called the offense merely “adequate” despite seven scores in seven series in the first quarter. That’s quite similar to what he was saying to the media after his NFL debut about the team playing too slow. No matter what, he’s always pushing for a faster and faster offensive pace.

College controversy

Yet, of course, Chip Kelly’s high-profile success wasn’t without some varied controversy. The negatives began mostly with Yahoo! Sports investigative report in March 2011 that shared how the program had paid over $28,000 for an eventually unsuccessful recruiting pitch. The Yahoo! crew followed up with more details that July.

A shadow then lasted over the final years of Kelly’s college days. Many felt it was just a matter of time before he made the more profitable leap to the NFL. A December 2012 Willamette Weekly article shared how boosters were never happy with his presence on campus. Kelly was never much into the pomp and circumstance required of a collegiate head coach: Fundraising, public appearances, socializing in general, etc.

Kelly made his leap for the NFL in January. Then in late June, over 27 months after the initial report, Oregon learned the penalties of the NCAA investigation: No bowl ban, three years probation, only one fewer scholarship and an inconsequential 18-month show-cause penalty for Kelly. It was considered a win for the university, and likely won’t ever affect Kelly again, but still will casually hang over his legacy such as Belichick’s Spygate.

Existing NFL influence

Here’s the part where many analysts diverged in discussing the “blur” offense creator’s future impact in the NFL: Was it too gimmicky? In a short synopsis: Absolutely not and it already existed, in pieces, in the NFL’s innovative offensive schemes of the last few years.

Chris Brown had two articles in 2012 about this exact topic — one at Smart Football alone in April and another at Grantland via Smart Football in November. In the first one, he shared some history of the no-huddle offense and that it was a “no-brainer” for it to explode soon in the NFL. Colleges were being far more innovative in this regard, a topic he featured more in his Chip Kelly profile over at Grantland. This article shared how the offense was “more familiar than it seems” with only a few basic formations and that Kelly’s NFL success was eventually inevitable.

The Boston Globe’s Greg A. Bedard wrote in October 2012 how the Patriots were using just one word to spark their no-huddle formations led by quarterback Tom Brady. Who else could have been the innovator of this? Kelly, who was called a “genius” by an NFL veteran. Again, Kelly’s footprint was all over Belichick’s historic success, a coach who was one of many to eventually seek out Kelly’s advice.

Chip Kelly’s future

Those fourth down calls epitomize Kelly’s aggressiveness but what the average football fan doesn’t realize is that Chip’s play-calls (the fourth down tries, fake punts, two-point conversions, etc.) are almost always the correct mathematical decision. Like Paul DePodesta and Billy Beane did in baseball, Kelly’s genius comes from exploiting arithmetic that other coaches are too naïve to acknowledge.

Tim Livingston’s November article for The Post Game sparked my initial math-related interest in Kelly. In it, more mathematical systems are introduced and expanded upon, the ones I shared being used for an Arkansas high school earlier. Much like what the A’s, Bill James and Baseball Prospectus did in the baseball world, this mathematical program estimated the impact every possible football play had on the probability of a win for a given team.

For many, the Chip Kelly stereotype featured a gimmick offense revolving around a mobile QB and a faster-paced version of the spread offense. But Livingston showed this was not the case. It was all based on math and win probabilities. My brother Adam was the one who intitially forwarded over that Livingston article. Knowing me so well and my affinity for math and Moneyball-esque revolutions, Adam’s comment was: “His math will revolutionize the NFL… hopefully for the Browns!”

That didn’t happen. In the end, Chip Kelly decided upon the Eagles and not the Browns. For one, it might be easier to point out how the personnel was likely a deciding factor. Running back LeSean McCoy is a more shifty and explosive back, like the Oregon sprintsters, as opposed to workhorse Trent Richardson. Mike Vick has proven to be an NFL star and the Eagles defense, despite its poor 2012 performance, also had proven in the past to be somewhat successful. This was going to be Kelly’s show and the players seemed to promote more early success.

While I think the Eagles will eventually fade to a mediocre yet exciting 7-9 or 8-8 this season, it will only be the catapult for a long track record of NFL success for Chip Kelly. It’s just bound to happen. Peter King noted at The MMQB how the Eagles’ Week 1 numbers did indeed mirror the Oregon numbers I shared above. It’s all going according to script. And most importantly for Chip Kelly, that began with a win, too.

  • The_Real_Shamrock

    Pretty easy to see why Kelly chose Philadelphia they have a QB, RB and WR who fit perfectly into his offense.

  • Ezzie Goldish

    Great piece.

    While the way Kelly seems to handle things made me not mind as much, and that Chud also does seem somewhat more math-inclined than most helps (as does his ability to adjust schemes), I was also saddened that we wouldn’t see the coach that will change the way the game is played here.

    More and more, I’m torn by the Banner & Co. philosophy, which seems both interested in being innovative and yet simultaneously stuck on certain philosophies (even if they’re generally successful, like Norv’s O or Horton’s D). That they clearly wanted Kelly makes me feel better about their mindset, though.

    That all said, it was interesting both that Kelly let up and that their O struggled after halftime. Did they let up because they were struggling? Or because they wanted to burn clock? How quickly will NFL D’s adjust? How quickly will the Philly D adjust to being back on the field so quickly each time, with less time to analyze the opposing O or catch a break and will they also have the stamina needed?

  • mgbode

    I was excited about the possibility of Kelly for just this type of thing. The attempt to get back to our roots and get a real innovator in charge of our football operations. It is a shame it didn’t work out that way.

    that said, the 2nd half left the door open for some doubts. Wash adjusted to the trap zone-read plays that were so effective in the 1st half and the yardage stalled. the drives that scored points tended to be on short fields.

    i think Kelly will be successful in the NFL, but, like you mentioned, I don’t think it is going to necessarily take full effect this season.

  • Tron

    Wow, Philly has a stud running back and Kelly chose to run the ball 62% of the time? That is smart, too bad Chud didn’t get the memo. Instead he chose the Manning/Brady offense without Manning or Brady…..

  • nj0

    I remember when the Dolphins were going revolutionize the NFL with the Wildcat. Now, it’s a sometime gimmick. Time will show how the league adapts.

  • nj0

    Laugh all you want, but Andy Reid regularly had about a 3:1 pass-to-run ratio in Philadelphia.

  • nj0

    I’m not a stats-head and only have dabbled in sabermetrics. My issues with advanced stats in the NFL: 1.) sample size and 2.) the massive amount of variables.

    In baseball, you have an amazingly detailed data set from a static game dating back nearly a hundred year. In football, you have far less data that is far less deep and which includes a massive amount of variables that create a lot of static. Maybe I’m just not hip, but I don’t see how these supposedly advanced statistics are going to change the NFL.

  • WFNYJacob

    As I wrote in the article, no NFL team has had a run ratio greater than 54% in the last three years. So 62% is pretty outrageous in that context. Game 1 for the Eagles: 63.4%.

  • WFNYJacob

    Philadelphia’s running play percentage last six seasons:
    2012 — 38.3% (25th)
    2011 — 43.3% (14th)
    2010 — 40.9% (24th)
    2009 — 38.5% (28th)
    2008 — 39.7% (27th)
    2007 — 40.2% (23rd)

  • WFNYJacob

    I think you’re doubting the sample size. Just like in baseball, every situation can be documented and probabilities measured. The value of the Zeus program — the mathematical computerized system that I mentioned in the article — is it puts all these probabilities in one place.

  • WFNYJacob

    He just needs more talent, I’d say. The Eagles were not a good team last year. They have some talent leftover from their good years, but it will take some time. Chip Kelly’s system alone will help them be exciting and mediocre, but not competitive just yet.

  • nj0

    Documented how? How rich is the metadata? Does it track personnel on the field? The placement of every player? If so, how far back does it go?

    I just think there are too many moving parts in football and too small of a sample to get anything all that useful.

  • WFNYJacob

    Philadelphia had just 24 plays in the second half. There were 19 runs and 5 pass attempts. Only 2 of those passing plays were completed and only for 13 yards.

    So yeah, they weren’t moving the ball in the air in the second half one bit. They also weren’t attempting passes at all, even at the rate they did in the first half.

    But, the Eagles run attack still averaged a sensational 5.89 yards per carry on those 19 attempts. McCoy gained 69 on 11 carries with a TD. Vick had 44 on 5 carries.

    So to say that the Redskins “figured out” the Eagles running attack in the second half is exactly mathematical right. The biggest factor was just cutting their plays in half and not passing the ball hardly at all.

  • WFNYJacob

    I obviously don’t know the intricacies of the program. But the Livingston article at The Post Game and the 2008 Rivals article both talk about the software in quite more detail than my article. I think it’s incredibly valuable for the future. At worst, they’ve been measuring for a decade and are thinking about it in innovative ways.

  • mgbode

    TMQB(Easterbrook) has documented the Arkie HS program since they stopped kicking the ball (except in extreme situations). I am actually surprised that some Sun Belt or MAC school hasn’t at least attempted to duplicate it in college yet.

    I completely agree on the innovating thinking that it may develop. Where I think NJ is likely coming from is seeing how some people jump to conclusions based on the statistics and use it to disparage coaches w/o looking at specific game sequences (*cough*Barnwell*cough*).

    also, of note, is that teams like the Ravens have admitted to utilizing more statistical s/w tools to help them determine draft order, et cetera. it may not end up being the same as baseball & basketball, but people will figure out how to use some of it to their advantage (and likely have been for years now behind the scenes).

  • mgbode

    yes, time will show on it, but I see his offense being more K-gun than Wildcat. In that the concepts his offense is based in are easier to continually adapt and change to adjust to the defense counter-punches.

    with the K-gun (Bills w/ Kelly+Levy), it’s still being basically utilized by the NE Pats, Denver Broncos, and NYG to varying degrees. Just adjusted for the newer style defenses.

  • mgbode

    I think he needs to see the adjustments and also figure out what will work vs. what won’t work at this level. Sometimes, there isn’t a way of getting around needing time and experience.

  • nj0

    I’m sure there’s value there. I just think stats in baseball have drastically changed the way the value of players is viewed. I just don’t think stuff like “going for it more frequently on 4th down” will have the same impact on football that advanced statistical analysis have had on baseball. I’m probably being overly dismissive and should educate myself a bit more, but still. That’s where I’m at now.

  • nj0

    There’s also the question of wear and tear. 30+ carries for McCoy. And Vick got killed too.

  • nj0

    I’ve always been interested by the idea of a team game planning to never punt. If you knew you had four instead of three downs, how would/could you run your offense differently? Maybe bring back a grinding, three-yards-and-a-cloud-of-smoke run game?

  • maxfnmloans

    do you think though, that perhaps that was also a function of Kelly deciding to “dumb things down” a bit because they had a lead, that way he didn’t have to expose more of his offense to upcoming teams than need be?

  • maxfnmloans

    Boy am I glad we played the Eagles in the first week of the season last year as opposed to this year.

  • WFNYJacob

    Certainly seems logical, too.

  • mgbode

    definitely. Bryce Brown needs to be more prominently involved. also, can Foles run? regardless, Vick needs to be smarter about “how” he gets hit. he can slide or get out of bounds quicker. but, that is true regardless of the offense he runs.

  • mgbode

    or, take more chances for explosive plays knowing you have 25% more chances at one working.

  • hash

    I always liked the philosophy of pounding the run game to soften the defense, then strike with the aerial when their guard is down.

  • BuckeyeDawg

    Count me in as skeptical that this “gimmick” offense will work long term. My prediction is that defenses (good ones, anyway) figure this out by mid-season once they have some game tape to work with. There is a big difference between terrorizing slower, overmatched PAC-12 defenses every week and going up against much more talented, world class athletes week in and week out in the NFL. Seems to me that quite often when Oregon played equally talented teams with well coached defenses, they lost. (Stanford, Auburn, LSU, and OSU come to mind)

    I could very well be wrong, but I see this going the way of the wildcat.

  • nj0

    I was speaking game to game, but that’s still not as bad as I remember or would have expected. My fault for falling for the media narrative.

  • WFNYJacob

    Your points all are very valid. I just believe that Kelly would then react to the situation and adjust. Several NFL teams are already running varieties of this no-huddle offense. I don’t really see what’s gimmicky besides possibly the zone read-based offensive schemes that already were all the rage in 2012. Kelly’s innovation and math-based approach just fascinate me that I think he’ll have much more success than the average coach.

  • mgbode

    Oregon had it’s share of good wins too though.
    KState, Stanford (53pts in ’11, 52 in ’10), Wisconsin, Oklahoma State, Michigan, a good Utah team, some good USC, Oregon State teams.

    and, even those losses you mention were close ones. Really, what team doesn’t have a worse record when just looking at the best competition? (also, you didn’t mention Boise who really had Oregon’s number)

    ———————–

    now, you could very well be correct in that some of the elements will be discarded. i fully expect that part. it’s just that I think it could become the “running” version of what Denver, NYG, and NE run.

  • BuckeyeDawg

    Fair enough. I would be curious to know what Kelly’s record at Oregon was vs. top 25 teams. Because of the parity in the NFL, I would argue that coaching in the NFL is like coaching against a top 25 team (or top 31, as it were) every week in college football. I just think that the speed, conditioning, and physicality of the NFL from top to bottom will negate a lot of the advantages that his system had in the college game.

    But, I guess we’ll see. If it ends up working and becomes the next trend in football, it won’t bother me. I can’t deny that it’s a lot of fun to watch.

  • mgbode

    I was actually trying to compile such a list. The issue was that noone seemed to have end of year rankings. I had a hard time counting a #10ranked Cal team from September that went on to lose 7 games (one example).

    However, his top25 record is quite pristine. He did well against the Oregon State, Stanford, USC, UCLA, Arizona State teams whether they were ranked or not.

    I don’t think that matters all that much though. What matters now is that he has to figure out how to adjust and adjust and adjust. That is the way of the NFL. I agree that “we’ll see” is as far as anyone can confidently go right now.