[[EDIT: [Note: At the urging of TwistNHook, this analysis post
doesn’t have now has pictures. Instead Furthermore, it talks about numbers, ratios, algebra, and equations. Please give it a chance as I think it is well worth the time and thought.] ]]
Run? Or Pass? That is the question.
For a long time I’ve been saying that a team’s rush attempts should be fairly equal to a team’s passing attempts. And, it makes sense. By keeping a balanced run/pass ratio of 1:1, the other team’s defense has no idea what to expect on any given play. Tedford has traditionally had about a 60:40 run pass ratio pre-Dunbar. In the previous years, Tedford’s playcalling has been about 60:40 run:pass in 2005, 60:40 run:pass in 2004, 55:45 run:pass in 2003, and 48:52 run:pass in 2002. As you can see, as Tedford’s tenure at Cal has gone on, he has moved towards a slightly run-first offense - the approximate 60:40 run:pass years (2003-2005). The year 2002 was somewhat of an abberation. Why? Because that year JT was rolling the dice a lot. The team wasn’t too talented, it was behind a lot, and had to pass a lot more. Interestingly, when Dunbar arrived in 2006, the run/pass ratio was about 51:49. This sort of makes sense too because Dunbar was a spread guy who passed a lot when he was the OC at Northwestern. So bringing Dunbar into Tedford’s offense dropped the run/pass ratio from 60:40 down to about 50:50. Did the equal run:pass ratio make the offense better? Well, aside from the Tennessee game in 2006, I think most people would agree that the 2006 offense was pretty darn good.
But let me pose a question to you that a friend and I discussed a while ago. Which offense was better? The 2004 Cal offense or the 2006 Cal offense? Here are the 2004 Cal Football stats, and here are the 2006 Cal Football stats.
You can make the decision yourself, but I give you my opinion: I think the 2004 Cal offense was better than the 2006 Cal offense. Case in point: greater yards per pass attempt, greater yards per rush attempt, greater yards per play, and more points per game.
But notice, the 2004 Cal offense’s run:pass ratio was 60:40 and the 2006 Cal offense’s run:pass ratio was 51:49. The comparison of these two years, the two best Cal offenses, goes contrary to my theory. Remember, my theory is that a balanced 50:50 run:pass ratio is theoretically superior than a biased run:pass ratio. What gives?
Well, I suppose my original theory of equal runs to passes was a bit of a naive statement. Afterall, Texas Tech and Hawaii pass the ball a lot more than they rush the ball and yet they seem to do just fine on offense. And teams like Air Force rush the ball a lot more than they pass the ball and they seem to do just fine on offense too. What gives?
Let me illustrate an anecdote that illustrates where my thinking went wrong.
But first I have to get something out of the way. What’s the best way to determine a team’s rushing efficiency? Yards per rush attempt (duh). But what about the best way to determine a team’s passing efficiency? Is it the QB rating? Yards per completion? Yards per attempt? Completion percentage? Obviously the question gets a little murkier when it comes to measuring a team’s passing ability. But I think the correct answer that most football analysts and coaches agree on is yards per attempt. Why? Well, yards per attempt essentially incorporates completion percentage, and yards per completion. Thus teams with a high completion percentage will have a higher yards per attempt, as well as teams who are better at passing (longer passing plays) will have a higher yards per attempt average too. The QB rating system is flawed because it punishes the QB for drops and interceptions which aren’t the QB’s fault, it punishes the QB for thrown-away passes which shouldn’t count, and gives the QB bonuses for YAC (yards after catch) of which he isn’t entirely responsible. But most of all it doesn’t measure how many yards are gained per pass attempt. And we need both systems of measurement for rushing and passing to be in the same units (yards per attempt) so we can compare the two. So yards per pass attempt is the best way to measure an offense’s passing efficiency.
Now, back to that anecdote that proves my theory wrong (my theory of having equal rush attempts as pass attempts is superior to a biased offense). Let’s just say there is a team with a perfect 50:50 run:pass ratio. Let’s just say they average 12 yards per pass attempt and 3 yards per rush. Does it make sense that the coach keeps his run:pass ratio at 50:50? No. It does not. If the pass is working better than the run, then you obviously should pass more (in general and excluding special situations such as goalline, two-minute drill, and burning the clock). This makes sense, doesn’t it? Why keep rushing just as much as passing when you are getting 9 more yards per attempt on passes than rushes? (Now, I know the argument is that an offense will get predictable if it deviates from a 50:50 run:pass ratio, but the counter argument to that is: to hell with predictability if the other team knows you’re going to pass (or run) and they can’t stop you!)
So when a team is extraordinarily better at either rushing or passing than the other, it makes sense for that team to do what they are better at even though their run:pass ratio will not be equal.
It is situations like these where I realized that my notion of equal runs to passes was wrong. In fact, while doing a little research, I did find that there is another school of thought that says exactly this. The gist is this: how many times a team should run or pass should be determined by their yards per rush attempt versus yards per pass attempt. The team should aim to have a slightly greater yards per pass attempt as yards per rush attempt. Why a "slightly greater" yards per pass attempt than yards per rush attempt? Because passing is inherently more dangerous due to sacks, fumbles, and interceptions. Thus in terms of expected value, a team should incorporate the higher risk of loss from passing by having a higher passing yards per attempt than rushing yards per attempt. How much more is debatable and indeterminate but we shall say probably around 1 to 2 yards more per pass attempt than per rush attempt. So in algebra terms, a team would want somewhere around X yards per rush attempt and X+1 yards per pass attempt or X+2 yards per pass attempt. For example: a team would want something like 5 yards per rush attempt and 6 to 7 yards per pass attempt. A team which has a 5 yard per rush attempt and a 10 yards per pass attempt is holding themselves back by not passing more and rushing less. Thus the team should pass more until the defense begins to compensate and the yards per pass attempt reduces to about a yard or so over the yards per rush attempt.
Another thing to remember is that by passing a lot, the rushing game opens up. And by rushing a lot the passing game opens up. So in the previous example where the team had a 12 yards per pass attempt average and a 3 yards per rush attempt average, if the team passed more and ran less, say around 30:70 run:pass, even though the team is rushing less, their rushing average should go up because the defense will compensate to defend against the more frequent passing (thus opening up the rushing game). So essentially what we strive to obtain, or what is ideal, is a yards per pass attempt that is slightly more than a yards per rush attempt regardless of run:pass ratio because it’s the offense’s effectiveness at the run or pass which dictates defensive reaction moreso than the offense’s tendencies. Let’s think this through. For example, let’s just say a team runs the ball 90% of the time but only averages 1 yard per rush attempt and the other 10% of the time that the team passes and averages 25 yards per pass attempt. What’s the defense going to do? The traditional notion of defending against the tendency (meaning defending against the run) would seem illogical here because the offense’s running game is so bad that it can’t get a first down whereas its passing game is more deadly. So the defense will defend the against the pass because the offense averages more yards per pass attempt than rush attempt even though the offense is more likely to run. So as you can see, I do think it is the offense’s effectiveness at the running or passing game which truly dictates how the defense plays moreso than tendencies.
Reviewing the Playcalling of the Tedford Years
Now, let’s look at some of the previous years for example in evaluating whether Cal should have ran or passed more.
In 2002, Cal had an abysmal 3.3 yards per rush attempt, and an okay 6.9 yards per pass attempt (stats available here). While the team did have an unusual (for Tedford) 48:52 run:pass ratio -and it is unusual because Cal passed more than it ran- statistics say Cal should have passed even more than it ran. Anybody remember Vinny Strang? And Jesus-in-Cleats Kyle Boller? Cal should have passed the rock more in ‘02, somewhere until the yards per pass attempt went down to about a yard more than the yards per rush. My guesstimation would be that it probably would have been somewhere around 5.5 yards per pass and 4.5 yards per rush attempt.
In 2003, Cal had a 4.3 yards per rush average, and an 8.2 yards per pass attempt average (stats found here). The team’s run:pass ratio was 55:45 run:pass. So Cal should have passed a significantly greater amount of time than they rushed. The greater passing would have taken advantage of the superior passing game and opened things up more for the rush. My guestimation would be that Cal probably should have had a 45:55 run:pass ratio, thus probably bringing the averages for rushing up to around 5 to 5.5 yards per attempt, and the yards per pass attempt down to 6.5 to 7 yards per attempt.
In 2004, Cal had an amazing 6.1 yards per rush average, and an 8.5 yards per pass attempt (stats found here). Those averages are INCREDIBLE. Both are great. The playcalling of this year was 60:40 run:pass but statistically could have benefitted from a little more passing than rushing. A run:pass ratio of 55:45 might have been slightly more efficient of the offense’s abilities.
In 2005, Cal had a 5.8 yards per rush average, and a 7.2 yards per pass attempt average (stats found here). The run:pass ratio of this year was 60:40 run:pass. The amount of yards per pass attempt is 1.4 yards per attempt more than the yards per rushing attempt. This is just about perfect. And thus this year was probably the best proportion of run to pass that we’ve seen so far.
In 2006, Cal had a 4.9 yards per rush attempt, and an 8.0 yards per pass attempt average (stats found here). The run:pass ratio for this year was 51:49. Dunbar probably should have called more pass plays this year to bring the run:pass ratio to around 40:60. Now this year is quite interesting. Especially when compared to 2005. Marshawn Lynch was Cal’s primary runningback both years. Yet the averages for Cal’s rush attempts went down from 5.8 to 4.9. What gives? Well, obviously OL differences but primarily Marshawn Lynch and his Heisman campaign. Money entered this year very highly touted as a runner. Defenses concentrated very hard on stopping him and played the run more than the pass thus opening things up for the passing game. This led to a lower yards per rush in 2006 compared to 2005, and a higher yards per pass attempt in 2006 compared to 2005. Dunbar should have taken advantage of defenses keying in on Money and passed the rock more.
Finally, in 2007 Cal had a 4.9 yards per rush average, and a 7.1 yards per pass attempt average (stats found here). The run:pass ratio was almost precisely 50:50. Despite this equality in running plays and passing plays, statistics say Tedford should have called a few more passing plays than rushing plays. Not a lot though. Bringing the run:pass ratio to around 45:55 probably would have made the most efficient use of the offense’s abilities.
Run less with Marshawn?! I must be crazy!!!
Now, I know I’ve been going over entire years to evaluate if Cal should have passed or ran more, but this same analysis can be applied to individual games. In fact, coaches should be applying these principles to every game. Afterall, every opponent and its defense is different. Coaches definitely should be looking at the yards per rush and yards per pass attempts during halftime to see what’s working and what’s not, and making adjustments accordingly. So let’s look at one game this past year (2007) to apply this theory.
Well, let’s look at the bowl game. The statistics show that in the 2007 Armed Forces Bowl Game against Air Force, Cal averaged 5.9 yards per rush attempt, and 11.3 yards per pass attempt (5.4 yard difference). The run:pass ratio was approximately 56:44 run:pass. So applying this theory of playcalling, Cal should have passed A LOT more in the bowl game to take advantage of the porous Air Force passing defense. I’m just guestimating, but a run:pass ratio around 45:55 or maybe even 40:60 run to pass might have been even better. Cal had 507 total offensive yards that day. Imagine how many Cal might have had if the playcalling had been more efficient. Maybe around 550 to 600 total offensive yards.
Here are a few more quickies:
*In 2007 when Cal played Colorado State, Cal averaged 7.9 yards per rush attempt, and 5.0 yards per pass attempt. Cal should have ran the ball more to take advantage of its superior rushing game and open things up for the pass on the subsequent fewer pass attempts. Colorado State on the other hand averaged 3.7 yards per rush attempt and 10 yards per pass attempt (CSU should have passed more).
*In 2007 when Cal played Arizona, Arizona gained 1.1 yards per rush attempt, and 5.0 yards per pass attempt. Arizona should have passed more (that game Arizona attempted 62 passes and 20 rushes).
*In 2007 when Cal played UCLA, Cal gained 2.2 yards per rush attempt and 6.8 yards per pass attempt. Cal should have passed more (that game Cal attempted 34 passes and 30 rushes).
*In 2007 when Cal played USC, USC averaged 5.0 yards per rush attempt and 6.5 yards per pass attempt - a perfect mix of run to pass. USC ran the ball 48 times and passed 20 times that game.
*In the 2007 Big Game, Cal averaged 4.5 yards per rush attempt and 5.4 yards per pass attempt - practically a perfect ratio of run to pass attempts. On the day, Cal attempted 47 passes and 24 rushes.
For TwistNHook, there was a benefit for playing UCLA.
So what this theory of playcalling does is redefine "balance." No longer does the word mean equal rush attempts as pass attempts. But near equal yards per rush attempt as yards per pass attempt (with slightly greater yards per pass attempt than per rush attempt because passing is inherently more risky than rushing).
So are coaches who still aim for equality in runs and passes despite glaring differences in the production of the two behind the times? Or is this new theory of playcalling nothing but bat guano? The following are a few anecdotes to help you decide.
*In 2007, LSU (12-2) averaged 4.9 yards per rush attempt, and 7.1 yards per pass attempt (2.2 yards more per pass attempt than rush attempt). LSU had 612 rush attempts and 442 pass attempts on the year.
*In 2007, Georgia (11-2) averaged 4.5 yards per rush attempt, and 7.1 yards per pass attempt (2.6 yards more per pass attempt than rush attempt). UGA had 509 rush attempts and 365 pass attempts on the year.
*In 2007, USC (11-2) averaged 5.0 yards per rush attempt, and 6.7 yards per pass attempt (1.7 yards more per pass attempt than rush attempt). USC had 517 rush attempts and 416 pass attempts on the year.
*In 2007, West Virginia (11-2) averaged 6.2 yards per rush attempt, and 7.8 yards per pass attempt (1.6 yards more per pass attempt than rush attempt). WVU had 628 rush attempts and 265 pass attempts on the year.
*In 2007, Air Force (9-4) averaged 5.4 yards per rush attempt, and 7.3 yards per pass attempt (1.9 yards more per pass attempt than rush attempt). AFA had 721 rush attempts and 214 pass attempts on the year.
*In 2007, Hawaii (12-1) averaged 3.4 yards per rush attempt, and 8.6 yards per pass attempt (5.2 yards more per pass attempt than rush attempt). UH had 279 rush attempts and 669 pass attempts on the year.
*In 2007, Texas Tech (9-4) averaged 3.4 yards per rush attempt, and 8.0 yards per pass attempt (4.6 yards more per pass attempt than rush attempt). Tech had 246 rush attempts and 763 pass attempts on the year.
*In 2007, Stanfurd (4-8) averaged 3.0 yards per rush attempt, and 6.0 yards per pass attempt (3.0 yards more per pass attempt than rush attempt). Stanfurd had 446 rush attempts and 424 pass attempts on the year.
*In 2007, Army (3-9) averaged 2.6 yards per rush attempt, and 6.0 yards per pass attempt (3.4 yards more per pass attempt than rush attempt). Army had 400 rush attempts and 380 pass attempts on the year.
*In 2007, Idaho (1-11) averaged 4.0 yards per rush attempt, and 5.9 yards per pass attempt (1.9 yards more per pass attempt than rush attempt). Idaho had 474 rush attempts and 392 pass attempts on the year.
Well, for top teams such as UGA, USC, WVU, and LSU, they averaged about 1.5 to 2.5 yards more per pass attempt than per rush attempt. So maybe the amount that the pass attempt average should be greater than the rush attempt average is actually 1.5 to 2.5 yards (instead of 1 to 2 yards).
For teams that run a lot, such as Air Force, their playcalling was pretty balanced actually despite the fact that they ran the ball 77% of the time. So having a "balanced" run:pass ratio can be acheived even by rushing a lot more.
And as for teams that pass a lot, such as Hawaii and Texas Tech, their playcalling needed more passes despite the fact that they already pass a lot (because there was a 4.6 and 5.2 yard per attempt difference between pass attempts and rush attempts). So as good as both of these teams were, maybe they weren’t as good as they could have been since their playcalling was not as balanced as it should have been.
Now for bad teams such as Stanfurd and Army, who both had fairly balanced run to pass ratios of roughly 50:50, they still sucked. Why? Well, they both averaged 3.0 & 3.4 yards more per pass attempt than rush attempt. Maybe they should have passed more and would have picked up an extra W here and there.
But even for teams who have a "balanced" offense and are a model of perfection, such as Idaho, you can still be really really bad and lose lots of games. I think the difference in Idaho as opposed to the top teams such as USC and LSU, is that (1) those top teams have better defenses, and (2) those teams while also having balanced playcalling, just averaged more yards per attempt because of having better players.
So is this theory of playcalling legit? I think so. It’s at least something to really consider. Just ask yourself: do you really want to keep a balanced run:pass ratio when you get say… 7 more yards per pass attempt than rushing attempt?
"I had a perfectly balanced offense in 2007. How come we didn’t go undefeated? Oh well. We’re still the undisputed 2003 National Champions, and the undisputed 2006 Pac-10 Champions, just like OJ didn’t kill Nicole, and just like I didn’t know that Reggie was illegally taking benefits while at USC, and just like my fan base is full of nothing but humble and friendly fans."