So, I set out on a quest to find out just that. What follows is not trade analysis. I am not going to tell you who wins or loses a trade less than a week after it was made. What we'll be looking at is what effect, if any, a strong passing game has on the running game and vice-versa.
So for this study, I gathered as much data as possible from 2005 to 2012.
In Part 1, we will focus entirely on conventional (non-advanced) stats. I chose two stats for this portion of the program: passing yards-per-attempt and rushing yards-per-attempt.
Tables for 2005:
Thoughts, notes and observations for 2005:
- Nothing too notable here. The top passing attacks had a slightly above average rushing YPA and the bottom passing attacks had a slightly below average rushing YPA.
- Meanwhile... both the top and bottom rushing attacks had slightly above average passing attacks? Being stuck in the middle stinks.
- The best passing attack in the league yielded the 12th-best rushing attack. The worst passing attack in the league yielded the 7th-best rushing attack.
- The best rushing attack in the league yielded the 21st-best passing attack. But Michael Vick was the QB.... The worst rushing attack in the league yielded the 12th-best passing attack.
Tables for 2006:
Thoughts on 2006:
- Up is down and down is up! Here in opposite land: The best passing AND rushing attacks yielded slightly below average rushing and passing attacks
- While the worst passing and rushing attacks yielded slightly above average rushing and passing attacks.
- Seriously, you can't explain that!
- The best passing attack in the league had the 15th-best rushing attack, while the worst passing attack in the league had the 24th-best passing attack. Here's a reminder: Jon Gruden ruined Tampa Bay.
- The best rushing attack in the league had the 23rd-best passing attack. But again, Michael Vick. The worst rushing attack in the league had the 10th-best passing attack. Kurt Warner was good.
Tables for 2007:
Thoughts on 2007:
- Tom Brady loves Randy Moss.
- Adrian Peterson is really good. This is his rookie year. His QB was Tavaris Jackson. Yep.
- 2007 should also be renamed: the year when every horrible player inexplicably had a career year. It's so weird. One good year in a sea of poop.
- The most notable thing here? The best rushing attacks gave a noticable uptick to passing attacks, moving those passing attacks .3ypa above average.
- The best passing offense in the league had the 12th-best rushing offense. While the worst passing offense in the league had the 14th-best passing offense.
- The best rushing offense in the league had Tavaris Jackson. The worst rushing offense in the league had the 21st-ranked passing offense.
Tables for 2008:
Thoughts on 2008:
- Not too much out of the ordinary here. Most years we see slight movement (.1) above or below average. This year, the biggest difference is seen with the worst rushing offenses, which produced passing offenses .2ypa below average.
- The best passing offense had the 19th-ranked rushing offense, while the worst passing offense had the 26th-ranked rushing offense.
- The best rushing offense had the 18th-ranked passing offense, while the worst rushing offense had the 12th-best passing offense.
- 2008 was the year Peyton Manning played on one leg. People forget that the offensive line was also in tatters (Jeff Saturday missed time). The running game was useless. But Manning still provided the Colts with a legitimate passing attack. While compiling these stats, the biggest take away, with regards to Peyton Manning, was just how consistent he was. Yes, he's one of the best, but what makes him THE BEST, in my mind, is that he was one of the best, year-in and year-out. So hard to do.
Tables for 2009:
Thoughts on 2009:
- So here's the first realy interesting thing I noticed: this was the year the Vikings acquired Brett Favre. Favre gave the Vikings an elite passing offense (4th). This was also the year that Adrian Peterson and the Vikings had their worst rushing offense. It's not a matter of losing carries, this was a YPA study. It's really weird. I guess I would have to look closer at the Vikings, maybe there were some injuries there.
- The best passing and rushing offenses produced slightly below average rushing and passing offenses. The worst passing and rushing offenses produced average rushing and passing offenses.
- The best passing offense had the worst rushing offense, while the worst passing offense had the 15th-ranked rushing offense.
- The best rushing offense produced the 21st-ranked passing offense. This is Chris Johnson and Kerry Collins. Another rookie RB producing elite numbers despite bad QB play
Tables for 2010:
Thoughts on 2010:
- Some logical results! Slightly above average corresponding production from the best offenses, slightly worse corresponding production from the worst offenses.
- We see the Colts continue to struggle running the ball. This is the first year we've seen real slippage from the passing game (at least via conventional stats). Also Manning's last year (on the field) in Indianapolis
- The best passing offense produced the 22nd-ranked rushing offense. The worst passing offense produced the 12th-ranked rushing attack.
- The best rushing offense produced the 6th-ranked passing offense. The worst rushing offense produced the 20th-best passing offense.
Tables for 2011:
Thoughts on 2011:
- LOOK AT THE PACKERS YPA!!! The league average in passing ypa takes a jump this year, as well.
- 3 of 4 cases keep with the trend: within .1 of league average. The only "outlier"? The top-10 rushing attacks pushed their corresponding passing games .2ypa above average.
- Philip Rivers, while a jerkface, deserves some love: San Diego consistently had one of the best passing attacks while he was the QB. But while Rivers, Gates, and Tomlinson got all of the headlines, maybe Vincent Jackson was the MVP of that team? The team's ypa remained strong through injuries to Gates and the decline and eventual departure of Tomlinson, but it was Jackson's exit at the end of 2011 that sent the Chargers' offense flying off the map.
- The best passing offense in the league produced the 26th-best rushing attack, while the worst passing offense produced the 23rd-best rushing offense.
- The best rushing attack in the league produced the 8th-best passing offense in the league while the worst rushing attack produced the 3rd-best passing offense in the league.
Tables for 2012:
Thoughts on 2012:
- Again, that weird thing: both the best and worst passing offenses produced above average rushing attacks.
- Outside of that, the better rushing offenses produced better passing attacks and the worst rushing offenses produced below average passing attacks.
- One thing I really look forward to seeing: the top-4 passing offenses were all led by mobile quarterbacks. All 4 of those offenses had dominant running attacks. This is the wave of the future. This is why I've been begging the Colts to let Andrew Luck be Andrew Luck.
- The best passing offense had the 2nd-best rushing offense. The worst passing offense had the worst rushing offense. But it's Arizona. Does that really count?
- The best rusing offense had the 31st-best passing attack. Let me remind you: A RUNNING BACK SHOULD NEVER EVER EVER (EVER EVER) WIN THE NFL MVP AWARD. NEVER.
- The worst rushing attack... wait, didn't we just cover this? Arizona sucks.
Final Thoughts/Summary Tables:
- While 8 years provided an enormous amount of data, I'd still say it's a relatively small sample size. As I believe we're playing a different game now, I don't think we'd learn too much from going further into the past. Instead, I'll have to add on to the numbers in the coming years.
- Just to reiterate: this is not about the Trent Richardson trade. No data here and no statistics from Sunday will tell the story of that trade.
- There are a lot of variables I can't consider. Namely the play of the offensive lines. Most metrics will judge an O-line in run blocking by how well the running game performs. Well, is that the blocking or the running back? Probably a mixture of both.
- While there seems to be real discernible pattern, that may change: there was definitely a trend in the past 3 seasons that saw the top passing offenses have better rushing offenses AND the best rushing offenses having better-than-average passing offenses.
- In that same period, the bottom passing and rushing defenses both produced below average counterparts.
- We should also note, that while the average passing YPA started to slowly climb in 2009, the average rushing YPA has also slowly climbed. Basically, offenses are just OUT OF CONTROL.
- Fun with math: From 2005 to 2012, there were 80 top-10 offenses and 80 bottom-10 offenses (8 years x 10 teams = 80!). So if a Top-10 passing offense produced a top-10 rushing offense 30 times and a bottom-10 Pass offense produced a top-10 rushing offense 21 times, that means an average (11th through 22nd) passing offense produced a top-10 rushing offense 29 times. Averages: Top-10: 3.75 top rushing offenses per year, Middle-12: 3.625 top rushing offenses per year, Bottom-10: 2.625 top rushing offenses per year.
- This probably should have been another table. But seriously, I can't stand another table.
- Top-10 rushing offenses: 3.75 top passing offenses per year. Middle-12: 3.5 top passing offenses per year. Bottom-10: 2.75 top passing offenses per year.
- As you can see, there is definitely a difference, but it's about 3/4 of a team, per year.
Overall, I think if we learned anything, it's that good offenses are good offenses and bad offenses, by-and-large, are bad offenses. There are exceptions: Peyton Manning doesn't need your running game. Neither do Tom Brady and Aaron Rodgers. And Adrian Peterson doesn't need your passing game. Neither do Chris Johnson or Jamaal Charles. I don't think having a great passing offenses makes it significantly easier on your running game, but I simply don't have enough data.
Which brings me to this: these are just conventional stats. They are fine, they help us tell the beginning of a story, but they don't allow us to tell the whole story. As the title of this piece suggests, this is not a 1-Part endeavor. I'll be adding on to this every week. Next week we're going to take a look at advanced stats, specifically Football Outsider's DVOA and ProFootballFocus' signature stats. All-in-all, I expect there will be quite a few parts to this. I also want to look at the relationship between having a dominant wide receiver and the running game. I also want to look at the effect "star" power has on the offense. We generally know who the top-5 "names" at quarterback are each year. We generally know who the media, fans, and players consider to be the top-5 running backs in the league each year. Does having one of these "star" players have an impact on the offense?
After those three (advanced stats, wide receivers, "stars"), I also want to look at whether or not having a strong running game actually affects the opposing pass rush. This has been stated a few times in the past month, and I'm not sure if it's true or not. So we'll find out in a month or so.
I want to give a special thank you to Kyle Rodriguez for building those pretty tables for me, and to Meagan Terlep and my wife (who shall remain safely anonymous) for helping me gather some of the stats.
I really commend you for the outstanding amount of work you put in this post, Greg (thanks to the friendly pressure of other bloggers, apparently *ahem*). Like you, I find the phenomenon intellectually fascinating, and it´s great you´re trying to sort through so many possible factors.
Frankly, I don´t think using these stats is showing any kind of definitive trend, for all the reasons others have mentioned, and I have no trouble believing advanced stats tell an entirely different story. One thing I believe could help is, instead of merely comparing the yearly stats to the league average, adding the stats from all teams in two subsets (let´s say, 2005-2009 and 2010-2012) since 32 is too small a sample size, and then show the results in graphics based on standard deviation and quartiles, which could be perfect for just this kind of analysis.
Anyway, this is a really intriguing enquiry, and your willingness to just delve into it to have solid answers in hand contributes to this site being the haven of honest, smart thinking it is.
Damn, Greg, that's a shipload (dirty word dodge!) of work you're puttin' yourself in for. You sure you got the time, not to mention the sanity, to tackle all that number crunching?
Don't get me wrong; I'd love to see your analysis. It's just that dangers of recreational math cannot be understated. I once tried to grind through some simple blackbody calculation using an equation derived from Boltzmann's Law (seriously, it wasn't hard, it was just plug n chug) just to analyze whether some pseudoscience idiot was trying to baffle me with BS (he was). I made it, but the side effect was that during the grind I blacked out and only came to after many dead, cute furry animals laid about me in twitching agony. With children crying nearby.
No, I don't know where the children came from. I. Said. I. Blacked. Out.
Anyway, I have no lessons to offer here, other than that children cry easily and cute animals can leave the most uncute corpses... but really, Greg, think of the children before you engage in this endeavor. And the small animals. We're beggin' ya.
I would like to see some basic diagnostic statistics on that results table, because it's screaming out to me "no statistically-significant correlation".
Hmmm. Thank you for going through all the work to put this together. I look forward to the future parts of this series.
Bravo, Greg. Does this balance the budget and provide world peace? No way. But does it address and sift through data for a pretty common football talking point (so we're not just talking out of our butts in future)? Yes, indeed. Very useful.
I think it DOES suffer from small sample size. Because of that, outlier talents (Vick, who seriously affected the run game in ATL, Manning who does not "need" your run game until the post season, etc) will have a greater weight.
I also think a perfect data set would look something like Pats/Colts when their QBs went down (year before, the bad year, and year after--talk about a smaller sample size!) and similar years when a team's top-10 rush game was taken apart by injuries--how did that absence affect the other side of the offensive coin? Like the Edge injury year (Even though Edge ran for 661 and Don ran for 1,104 that year, we were not a good running team; Dom only got yards because of opponents' fear of 18.
It looks like there is some correlation to me--not guaranteed or ironclad, but it's there and that fits with the "good offenses are good and bad offenses are bad" thesis. Both can be affected by a terrible OL, or being in a division with three dominant defenses (a minimum of six brutal games for your O each season).
Basically, it supports my hunch that any player wants to play with other good--or even better--players because they will lift everyone else around them. Yes, as an RB on Peyton's team I might get fewer runs, but as a team we will do better and I might make up for lack of runs with more catches. And I will have higher efficiency grades for every touch. As a RB on team X I might expect 20 carries a game for 4.0 yards, but if that team suddenly lands a great QB or WR, I might expect only 15 carries for 5.0 YPC. Did I just get better? No, but somehow I am doing better (not by traditional, dumb, counting stats, of course). Will wonders never cease....
I am sorry to say but I am not sure how worth is the time taken to try figuring this thing out. Ultimately is about making the best decisions, run or pass. I don't believe in making "philosophical" decisions about being a passing or running team.
I think I may limit myself to enjoying the football games because this continuous battle between "football philosophers" is getting me tired. I don't mean to demean the people who enjoy this debate. I just don't. This is my favorite Colts related site and that's why I care enough to post this.
I get the feeling that the outliers will fly against common sense takeaways. By that I mean, with an average QB and an average running game, it seems obvious to think that having a credible run threat allows the play action to work over and over. And not having a credible run threat allows the defense to drop the safeties into coverage and shut down the offense.
Look at last night's game. The great camera angle over Peyton's shoulders showed him seeing the safeties cheat up, worried about the run and then he'd torch them with a pass. Conversely, when he saw the safeties cheat back for the pass, he'd pound the run over and over. Just carnage and really fun to watch.
However, having a really good QB and really good receivers may mean that it doesn't matter whether you have a credible run threat because you'll complete the passes no matter what. Your receivers will get open even on press coverage. This would be the "outlier" condition. I guess the reverse is an Adrian Peterson who runs so well it's very hard to stop him, so the Vikings will do well even without a credible passing game (though maybe that was disproved this weekend versus the Browns).
Anyway, just a thought.
@ColtsAuth_Kyle Thanks, Kyle. Checks in the mail.
@Lvl9LightSpell I don't blame you, I didn't read it, either.
Again really good write up, one reason why I love this site is because seems like people care about putting out rational information and don't just say whatever the heck they feel like just to get attention. Hopefully this site continues to stick around for many more years.
But if I happen to draw certain conclusions about the Richardson trade from this article, that's okay, right?
I love it. I agree that this isn't about the trade for Richardson, but a chance to look at the "system" as it effects QB play. I think that the advanced stats will show much more than the traditional ones.
Curious if there is a way to look at "Great QB's" and their running games. For example: How do guys we know are great passers stats change when they have a good running game or bad one? Do they not change at all? When Brady and Manning were younger, they both had better run support. Did it matter? What about Rodgers, Brees, Marino, Elway, Young, Montana, Moon? I've long held that being a great running team isn't necessary. However, being an efficient running team (over 4.0 ypc) gives your team a much higher margin for error than being a great QB with a poor running game. I can't wait to see how your research effects that perception.
Jokes aside: Thanks for the analysis, and I'd love to see what else you come up with. Regardless of the cute furry animal body count. :D
@ColtsHead_Ben What about post hoc correction for multiple comparisons? :)
@ColtsHead_Ben I agree with you on both accounts (didn't think it was statistically significant, believe they warrant further investigation). Due to the amount of data, I left it more open-ended, but when I come back around to tie this "series" up, I believe we'll need to "tie up those loose ends"
@bradicus18 You don't seem impressed. I will add another table.
@Bobman1 Hey Bob, first thanks for the kind words.
Your idea about looking at offenses that suffered the loss of a key player (Patriots/Colts with Brady/Manning) is really good. I know off the top of my head that the Colts game improved significantly, on a YPA basis, under Painter/Collins/Orlovsky.
The Patriots had the 7th-best offense by YPA during 2008, when Cassel was at the helm.
As for the rest of your post, there's a saying in hockey: skill needs skill. Sorry for the hockey analogy, but there's a guy who plays for the Penguins named James Neal. With an elite center (like Evgeni Malkin), he's a 40-goal, point-per-game player. With an average center, he's a nearly useless.
This doesn't make him a useless player, however, as playing WITH skill is a skill in itself. And scoring 40g is nothing to sneeze at. It doesn't matter if he needs the elite players around him, because it's not like a 40g player is common.
If, bringing it back to football, Trent Richardson were the same - not great with bad players, but capable of producing 5ypc and good receiving numbers with good+ skilled players - that would be great. Those yards still count. If he "needs" Andrew Luck to perform at that level, great, he HAS Andrew Luck.
It's going to be interesting to see what shows up moving forward.
@codrutc I totally get what you're saying. There are a lot of stats. Some might say an over-saturation of stats and analysis. At the end of the day, it's a game played once a week with a winner or loser. We spend 6 days a week breaking down the ins-and-outs of it. Sometimes we forget to just enjoy the game.
The majority of what I, personally, write is opinion-based editorials. I enjoy those the most. I like writing "pretty" things. This specific discussion interested me. It was something that I felt was... at least somewhat "approachable" from a research perspective. I didn't think it would be too hard to at least take a decently-thorough look at it.
Yea, it's a lot of data, but I think it's pretty easily-digested, if you can get passed the fact that there are 34 tables.
That said, it's not going to be for everyone. I get that, I respect it. But I think it's healthy, as a writer and as a site, to have a variety of topics. If you don't like this and don't think it's applicable or time-worthy, I get it, and we'll try to keep you engaged with the next article.
@buymymonkey Which is why Manning totally deserved the SB XLI MVP despite the numbskulls saying "he had ordinary numbers" and "Rhodes ran for 100 yards," etc. He conducted that game, in a downpour, like a maestro, taking what a very good Bears D (yes, missing Tommie Harris) gave him.
It's why baseball catchers deserve more recognition for no hitter/shutout games. There's a mental aspect to these games that is not obvious, especially from the stands (you only see the play once, at high speed, and can't focus on the small details from every angle), especially after a few beers (everything's really blurry, and oh crap, just got mustard on my hoodie....). TV is a pretty cool way to watch these things.
@buymymonkey yea, that's what I was getting at in the end of the piece when I was saying "good offenses are generally good and bad offenses are generally bad." I think they support each other. Although it does seem, just from the numbers, that the running game is more immune than the passing game. That is, the running game was able to stay closer to average with a bad passing game than vice versa.
@GregCowanCA Can we just do direct deposit. Paper checks are so 1986
@Kyle Rodriguez Someone who didn't want to see me finish this last night...
@paulcareyjr Well, I see somebody is hoping to get a free subscription! Well I paid my $100 (in cash) and that's fine with me.
What, there's no fee to join this site...? Uh-oh. Where the hell did I actually mail that c-note...?
@paulcareyjr thank you, Paul. We try.
@Heracleitus It's pretty open-ended. I'm certainly allowing people to draw their own conclusions, yes.
I just wanted people to understand that I didn't just go through collecting 34 tables worth of data to prove or disprove a point. I was interested in it. I wasn't looking to poop on the trade, I genuinely wanted to know.
If you took something from all of this, I've done my job.
@Music Man Yea I mention in the closing section that part 3 of this will be a "Stars" look.
I'll be looking at Star QBs and their running games and star RBs and their passing games.
I've talked to Kyle about this. If I have the time, there's about 10 different "correlations" I want to test. Lots of interesting angles to explore.
I honestly think THIS ONE was the hardest one, so I feel okay moving forward. We're see.
I have to be cognizant of the fact that some people just skip to the end. Also, this is no excuse, but when you're staring at 36 tables (the original number, Kyle compressed 2 tables for me), you kind of lose track of what you've done and everything bleeds together.
Finally, and most important: my high school English teacher is grading this, I had a word count to meet!
@GregC @bradicus18 No! No! I didn't want you to think that. I'm impressed with the lengths you went to put together this data. I'm just not sure we learn much from it other than, as you said, good offenses are good offenses and bad offenses are bad offenses. I just have questions, that's all. I'm not sure they are relevant questions and they might be answered in future posts.
I am curious, why top 10/bottom 10? Why not top 5/bottom 5 or top 7/bottom 7? Ten teams make up 31% of the league and bring the results closer to league average. Though, using top 5/bottom 5 may not tell us anything more except really good offensive teams are really good offensive teams and really bad offensive teams are really bad offensive teams.
I'm looking forward to future posts...hoping for MOAR TABLEZ!!!!
It seems that for the Richardson trade to eventually satisfy those who are currently opposed to it, the Colts are going to need to get some sort of x-factor bonus from it; something where 1 + 1 = >2. Like if the potential threat of Luck + Richardson becomes significantly more than either of them playing with a lesser QB/RB would have been (or was until now).
@ColtsAuth_Kyle harder to launder direct deposit.
Real world English 'teachers' know that good writing means lower word count.
All I did with this article was look at the two numbers in the bottom right of each table and repeatedly think "no statistical significance" while scrolling through.
@Pied I believe statistics is just that, statistics. I think what you are referring is what I would call "blind" versus "smart" application of statistical principles.
@GregC What is the difference between "advanced statistics" and just "statistics"? Thanks.
I feel like you wasted your time. Didn't really waste mine.
I think if someone like Aaron Schatz had done this he would have thought, "well that doesn't tell us anything, back to the drawing board" and not bothered posting it, or just summarizing the null result.
I'm with you in spirit tho. I look forward to your next iteration with advanced stats. Honest.
You're just too cool for school brosef. Advanced stats are coming. I'm going through the process, sorry you don't like it
Ya I think the moral here is 'conventional' stats are worthless and this kind of number crunching is a waste of time.
Stick to advanced stats please. We all have a chance of learning something that way.
It's an interesting discussion, for sure. My decision was based on two things (and maybe I should have just doubled up and did this with yards AND yards-per-attempt): the original comment that got me interested in this was about yards-per-attempt and Trent Richardson. So I wanted to keep it in that context.
On top of that, again, if a team wants to hand off 600 times a year, they can still rush for 2000 yards and not be a "good" rushing team. Just a persistent one. Does that have anything to do with their passing game? Or their coaching staff? (I'm looking in Jack Del Rio's general direction)
I don't know what else I can say. That's my thought process.
@GregC @bradicus18 I understand what you are saying. I just think that in many cases, a #8 passing attack is closer to average than it is to the top. A #8 passing attack is a little more respectable than the #16 attack but could actually be closer to that #16 than #1. This seems even more pronounced at the bottom(#32 vs #16).
Also, does YPA indicate a top rushing/passing attack? Do NFL defenses game plan for YPA, YPG or other metrics? Do they perceive a team that passes for 8.0 YPA and 189 YPG (Seattle, 2012) to be more of a passing threat than a team that passes for 7.6 YPA but 291 YPG (New England, 2012)? Which passing offense would you try harder to shut down?
If we are trying to determine the effect one part of the game has on the other, are we looking at the right stats? Although, I think you are already exploring that. And do the results we choose represent good and bad or closer to average? Hell, maybe running and passing offenses are mostly average and it all doesn't matter. My head hurts. I think I'm going to stop and go get a drink.
Seriously, thank you for your work. I'm not sure any of my questions are relevant. Probably not, but I'm curious.
I wasn't looking for "dominant" offenses as much as "good." The argument was: He couldn't be an above-average running back because the passing offense was inept!
So I, in my mind, I guess? I judge average as ~11-22. I mean, 15/16 is average, but they are so tightly-packed in that range that, is there really a difference between 4.1230019 and 4.1229999 that I can say the cut off is X. So I just said 1-10: above average (or better), 11-22 average, 23-32 below average (or worse).
I think limiting it to top-5 and bottom-5 would have told us something, but I think it would have told us a "different" something. Like: Just how reliant on a running game is a dominant passing attack? Or vice versa. Or, how likely is it to build a dominant X offense with a bad Y offense, etc...
I wasn't really looking for dominance as much as "good"
Does that make sense?
@pierrezombie @GregC @Bobman1 Exactly. Of course quantifying separately what happens on a football field is a titanic task, but if skill+skill is right, we should see Richardson bring things to the table far above those of a replacement-level player, and we should also see Luck gain access to some news possibilities, like when you´re leveling up in video games, and your character looks exactly the same but has more resources internally.