A Synthesis: Sacks And Pass Defense

Earlier I made a post about 2006 SEC pass defense, and also about 2006 SEC sacks, analyzing each separately.

In reality though, considering it is all one defense, the two things are related, or are they? That's basically what I set out to determine. At bottom, is there any correlation between the number of sacks a team generates and how well their pass defense performs? Or does that number of sacks created have little if any correlation to how well a pass defense performs, as Joe Kines contended?

To figure it out, I ran the numbers on how well Adjusted Sacks correlated to a variety of different measures of performance for a pass defense, including opposing quarterback's average passer rating, yards per completion, yards per attempt, completion percentage, interception rate, and touchdown rate.

Before I list the results of what I found, I'll give a very brief primer on correlation for those who may not be familiar with the concept. At bottom, correlation ranges from 1 to -1, and values that that are near 1 or -1 have correlation, and values near 0 essentially have no correlation. Going up to 1 means there is positive correlation, i.e. the more x happens, the more y happens. Going down to -1 means there is negative correlation, i.e. the more x happens, the less y happens.

So, with that in mind, the following is how well Adjusted Sacks correlated to that variety of measures, in descending order:

Opposing QB average passer rating: -.4997
Touchdown rate: -.4693
Completion percentage: -.4660
Yards per attempt: -.4489
Yards per completion: -.3087
Interception rate: -.2573

As you see, opposing team's passing numbers generally tend to go down greatly as the number of Adjusted Sacks increase.

Opposing quarterback average passer rating, touchdown rate, completion percentage, and yards per attempt all have a pretty high negative correlation with the number of Adjusted Sacks. Again, there's just no real way to dispute the data, for the most part. Unless it's just an odd-ball deal with a one year sample size, which I highly doubt given the breadth of the statistics, it generally just showcases that pressuring the quarterback pays off.

I was thinking, perhaps, that would I might find would be a good bit of teams that racked up a lot of Adjusted Sacks did so by blitzing heavily, and that would even things out in the end with that defense giving up a lot of big plays and completions. But, as the breakdown showed, that simply wasn't the case. Certainly some teams did get a high number of Adjusted Sacks by blitzing heavily, but they didn't seemingly give up very many big plays or lot of completions. I suppose it isn't quite like the old Joe Lee Dunn days at MSU, where two sacks would be followed by a 38-yard completion against a seven man rush. You have to give credit to defensive coordinators and other defensive coaches; generally, it seems, even when they bring lots of pressure, the defensive backfield still, relatively speaking, prevents big plays.

The only snafu to the entire analysis was Interception Rate. There was a -.2573 correlation between Adjusted Sacks and Interception Rate, meaning that as Adjusted Sacks went up, the Interception rate tended to go down. Perhaps this should have been too shocking, especially considering the result of LSU, Arkansas, and Alabama, but it still is just a bit. You would think, intuitively, that more pressure would result in more bad throws and more interceptions, but the reverse is true. I suppose that the man-to-man coverage on the outside, combined with the shorter throws when the blitz is expected, means that defensive backs are not in as good of a position to intercept the football. And moreover, again since they are in man-to-man coverage, cornerbacks may not want to be as aggressive, knowing that they would have no help over the top if they made the slightest mistake. I know, it doesn't seem right intuitively, but more sacks correlates with fewer interceptions, at least in this data set. I would point out, however, that a correlation of only .2573 isn't particularly high, so maybe it is just statistical white noise that would work itself with a larger sample size. I suppose we'll see when we run the same numbers this time next year.

At the end of the day, you just have to be brutally honest about it. At bottom, it seems that if you were going to be successful in pass defense, you had to get after the quarterback. If you didn't, you just weren't going to be very good in defending the pass.

As for Kines, I certainly don't want to criticize his philosophy, because obviously he's forgotten more about football than I will ever know, but the numbers just seem to invalidate his defensive philosophy. Kines essentially betted that we could drop seven and eight defenders into heavy zone coverage, and opposing quarterbacks wouldn't be able to beat it. Unfortunately, at least in 2006 (that philosophy worked with all of the talent we had in 2005), that didn't work. Quarterbacks did effectively pick apart the defense with a relatively high degree of regularity, and with that in mind we struggled greatly in terms of pass defense. And, honestly, a lot of that credit goes to opposing quarterbacks. Perhaps there was a time when you could count on opposing quarterbacks to make a lot of mistakes throwing into heavy zone coverage, but that time seems to have passed.

The good news is that with Saban in charge now, Adjusted Sacks will go way up, and that should drive opposing team's passing numbers considerably lower.

2006 SEC: Adjusted Defensive Sack Rates

We've already covered offensive sack rates to help determine who protected their quarterback the best in 2006, so we'll now turn to defensive sack rates to help determine which teams rushed the quarterback the best in 2006.

As was the case with total offensive sacks allowed, defensive sacks created on its own is a pretty meaningless number, and only takes on real meaning when you put it into the context of pass attempts. Long story short, a team that piles up a lot of sacks can be a relatively poor pass rushing team, and a team that has only a relatively few sacks can in fact be a good pass rushing team, depending on how many passes they have thrown against them.

So, what I have done is taken the total number of sacks created by a particular team in SEC play in 2006, and divided that by the total number of pass attempts against the same team. That division yields a percentage of passing plays that I call Adjusted Sack Rate, which of course is the percentage of passes that resulted in a sack.

The following is how the SEC ranked in terms of sacks in 2006:



Moreover, we'll break things down a bit further. I understand that Adjusted Sack Rate can be a bit hard to grasp, being a raw percentage and all, so I also added Adjusted Sacks. Basically, Adjusted Sacks is the number of sacks that a team would have gotten had they faced the league average of pass attempts (223). Again, it doesn't change the data at all, it just makes it a bit easier to comprehend and fully understand by putting it in concepts more familiar to football fans.

Also, I've looked at teams not just by sacks alone, but also the lost yardage from those sacks. As you'll see, I've ranked the SEC in terms of the average yards lost per sack as well.

Without further adieu...



So there it is. If you've ever wanted better statistics regarding sacks in 2007, now you have them.

Honestly, I really don't have a whole lot to say regarding the data itself, and the main reason I did this was to get data needed for a later post.

Why do you ask?

At bottom, a sack is a pretty meaningless thing, in and of itself. You often times see teams that have a lot of sacks because they consistently blitzed everything but the kitchen sink, and while that did accumulate a lot of sacks, it still did not mark solid defensive play. Opposing teams would know their heavy blitz tendencies, and game-plan accordingly with a lot of max protect packages and short drops, with the end result being lots of completions and many big plays when a defensive back, playing man-to-man coverage, would miss a tackle and there would be no one over the top to clean it up. So really the reason I did all this is to compare it (hopefully in a post coming shortly) to 2006 SEC pass defense results. Again, hopefully that post will be coming shortly.

But I will say a few things...

Namely, Alabama just didn't get it done. The pass rush was a little better than most would have thought (eleventh instead of dead last), but there wasn't much to it. As I've posted before, the underlying idea behind the Kines scheme with regard to pass defense was to drop a lot of defenders into coverage, usually only rushing three or four defenders, and then forcing the opposing quarterback to throw into heavy zone coverage. And, well, that's exactly what we did. It just didn't work. We didn't get very many sacks (as expected, even by Kines), but we also didn't force incompletions with the heavy zone coverage. Again, at bottom, it just didn't work. You really can't put it any way other than that.

Georgia does need to be mentioned. The Dawgs finished third in the conference in total sacks, which sounds about right, considering their two outstanding defensive ends Quentin Moses and Charles Johnson. However, once you analyze the Dawgs through Adjusted Sack Rate, they only finished fifth in the conference, and really just looked like a very mediocre pass rushing team. Honestly, that shocked me. I fully expected Georgia to be one of the, if not the, top pass rushing teams in the conference. With Moses and Johnson gone to the NFL, and Georgia having to replace eight starters on defense in 2007, perhaps there are some signs for concern in Athens. All told, the Dawgs were the textbook example of racking up a lot of sacks against the Sisters of the Poor, racking up ten sacks against the, shall we say, vaunted triumvirate of Western Kentucky, Colorado, and UAB (a 6-5 Division 1-AA opponent, and two D-1A opponents with a combined record of 5-19. If you ever wanted a case study of why I only look at conference games, this would be a good one.

The rest really aren't too shocking. The teams that you thought rushed the quarterback really well (LSU, Florida, etc.) in fact did so, and the teams that you thought rushed the quarterback poorly in fact did so, too.

The only other mild surprise, to me, was Tennessee. The Vols are always very talented in the front seven, and I expected them to be a pretty solid pass rushing team, and they were far from it. For whatever reason, the Vols struggled to get after the quarterback, finishing tenth in the SEC in Adjusted Sack Rate, and had only about two Adjusted Sacks more than Alabama and Ole Miss. Yikes for the Vols. Not what I expected.

Finally, let's look at average yards lost per sack. Arkansas, LSU, and Florida (i.e. three of the top four teams in Adjusted Sack Rate), finished 9th, 10th, and 12th, respectively, in terms of average yards lost per sack. At bottom, teams knew they had a great pass rush, and planned shorter drops accordingly. And, of course, shorter drops equals less yardage lost when in fact a sack does occur. On the other end of the spectrum, the two teams that finished first and second (Alabama and Tennessee) in average yards lost per sack were about the two worst pass rushing teams in the conference. Opponents saw the lack of a pass rush, and planned deeper drops and longer routes, accordingly. That, in turn, resulted in a higher amount of yards lost per sack, when in fact a sack did occur.

Later we'll put all of this together with 2006 pass defense and see what we find. It should be interesting stuff.

The Scoreability Index

A couple of days ago, I posted about the Bendability Index, and made the case for its high relevance in college football.

Today, we'll move from a statistic that focuses mainly on the defensive side of the ball to a statistic that focuses mainly on the offensive side of the ball: the Scoreability Index. Whereas the Bendability Index determines how many yards it took an opponent to accumulate to score a point on a team, the Scoreability Index determines how many yards it takes an offense to score a point on their opponent. Just like was the case with the Bendability Index, the Scoreability Index is not solely a measure of offensive prowess, but instead a measure of overall team efficiency that takes into account a variety of factors, such as red zone offense, special teams performance, defensive performance, and turnovers.

The following is how the SEC finished up in terms of the Scoreability Index in 2006:



Now let's look at things closer, beginning with my beloved Crimson Tide.

Unfortunately, well, it's just not pretty for those clad in crimson. As you can see, we finished eleventh in the conference in the Scoreability Index, and combined with us finishing eleventh in the Bendability Index, well, no one should be surprised that we went 6-7 and Mike Shula was fired. In terms of the Scoreability Index, the offense moved the ball relatively well. We racked up 2,691 yard of total offense -- granted, keep in mind I understand that looking at things by total yardage alone doesn't tell you much, which is why we're going farther in this analysis -- and moreover we were very consistent in racking up yardage. We weren't boom and bust (400 one week, 150 the next), we consistently put up over 300 yards almost each and every week (the only two exceptions were Tennessee and MSU). But unfortunately, those yards didn't really translate into very many points.

At bottom, though we moved the ball relatively well, it didn't translate into points. Poor special teams (all across the board, in terms of punt returns, kick returns, punting, kicking, and punt and kick coverage), and below average defensive play meant that we generally didn't have great field position, so we had to move the ball long distances to score. Beyond that, once we did move the ball downfield, we bogged down in the red zone, so we never saw the full potential point value of all of the yards that we had accumulated. For example, moving the ball 74 yards isn't very meaningful if you can only get three points on the board when it is all said and done.

At the end of the day, it was just bad from top-to-bottom. There are no real bright spots to the matter.

Now just a few thoughts on the rest of the SEC...

Vanderbilt was the only team that finished behind Alabama, and in reality we should have been behind even them. Vanderbilt's yards-per-point ratio was skewed massively by the fact that in one game (Kentucky) they put up 621 yards of offense. For the other seven games, we were a good bit worse than even the Commodores. Bad stuff.

Arkansas just set the world on fire in this category, needing only 11.5 yards per point. You have to give the Hogs credit, they were good from top to bottom. They played great defense, capitalized on turnovers, had good special teams, and ran all over just about everyone when they actually got the ball on offense. What can you say? They were the epitome of how it is supposed to be done.

Auburn did not have a very good offense in 2006, but they did quite well in the Scoreability Index, finishing third in the conference. Again, they weren't very good on offense, mainly due to injuries to Brandon Cox and Kenny Irons. However, they still did the things, as an entire team, needed to be successful. They got great special teams play, and great play from their defense. Though their offense wasn't particularly good, it was efficient and capitalized quite well off of what their special teams and defense created for them.

And finally, to close, one of the keys to always look at with any statistic is how well that particular statistic correlates to winning or losing football games. Some stats -- despite people talking about them all the time and you hearing it talked about on ESPN 24/7 -- have essentially no correlation to winning or losing football games, but that is not true of this statistic. Much like the Bendability Index, the Scoreability Index has a very high correlation to actual wins. Just look at the teams that finished in the top five in the Scoreability Index versus those who finished in the bottom five. The teams that finished in the top five of the Scoreability Index were a combined 54-13 (80.59 percent winning percentage). The teams that finished in the bottom five of the Scoreability Index were a combined 28-33 (45.9 percent winning percentage).

Small wonder Cold, Hard Football Facts -- the inventor of this statistic -- refers to it as one of the "Stats That Matter."

More On Pass Defense

In the past few weeks, I've made several posts on 2006 SEC pass defense. The problem with it all, however, is that its hard to put together. You analyze a ton of pieces, but at the end of the day it can be difficult to put them all together and determine what they mean as a whole.

But what we can do is take the combined statistics (attempts, completions, yards, touchdowns, and interceptions) of every SEC pass defense and convert them into one simple, easy-to-understand formula: the quarterback passer rating.

Granted, the quarterback passer rating has some flaws, but generally it presents a good overall summary of performance.

Thus, here you go, the 2006 SEC pass defenses ranked in terms of opposing quarterback's cumulative passer rating:



And that sums things up quite nicely.

As you can see, the Crimson Tide ranked ninth in the SEC in terms of opposing quarterback's cumulative passer rating. At bottom, we looked good on total pass defense (4th in the conference) because we allowed so few passing yards. But the dirty little secret behind that was we saw the fewest passes thrown against us of any team in the conference. In reality, we were a quite poor team in terms of pass defense, better than only lowly Mississippi State, Kentucky, and Vanderbilt.

2006 SEC: The Bendability Index

In the previous post, I introduced the Bendability Index, and made the argument for its relevance in college football. Though I focused on Alabama, I nevertheless had to do the research for the other eleven SEC teams just to see how the Crimson Tide stacked up by comparison. And since the research has already been done, I might as well go ahead and publish it.

The following is how the entire SEC ranked in Bendability Index:



All told, I wouldn't really say there are any major surprises. The teams generally stacked up about like you thought they would have, with a few exceptions here and there.

One point, though, that should be made regards Georgia. Though the Dawgs had one of the best defenses in the conference -- if not the nation -- in 2006, they were very low in terms of Bendability Index, ranking tenth in the conference by that measure. The problem was a struggling Georgia offense that often times committed devastating turnovers. Again, as said earlier, the Bendability Index is not solely a measure of defensive prowess, but of overall team efficiency. The 2006 Georgia Bulldogs proved that point quite nicely.

The Bendability Index

The Bendability Index is something you probably haven't heard of before. But it's something you are about to hear about, and it's something that you should know. It's a very relevant statistic, and it should be tracked, though -- for the usual reasons regarding a lack of in-depth analysis in most circles -- it is not.

So just what is the Bendability Index?

Cold, Hard Football Facts -- insofar as I can tell -- is the inventor of this statistic, and they have this to say about it:

This is the first stat that chronicles the phenomenon of the "bend but don't break" defense and provides a measure of defensive efficiency. The Bendability Index is obtained by dividing a team's total yards allowed by total points allowed, yielding Yards Per Point allowed. A team that ranks high on the Bendability Index has the defense that opponents must work hardest to score upon

In other words, you literally have to do more to score points against teams that rank high in the Bendability Index. You have to gain more yards, move the ball farther, pick up more first downs, etc.

But is the Bendability Index merely a measure of defensive prowess? No. Again quoting Cold, Hard Football Facts:

The Bendability Index is not purely a defensive yardstick. It is, instead, a great barometer of team success, of overall team strength and efficiency. It is a function of many team-wide factors, including general defensive strength, special teams proficiency, turnover differential, and Red Zone defense.

Though the Bendability Index measures defensive prowess, as noted earlier, it also indicates overall team strength. Poor play from the offensive and special teams (namely offensive turnovers, and poor punting and kicking) also have a massive impact on the number of points allowed. So it's not just a measure of defensive performance, but of the efficiency of your team as a whole. Never forget, it's not just defenses that give up points. Teams, as a whole, give up points.

Moreover, it's a highly relevant statistic. Double Extra Point -- a blog similar to this one dedicated to Nebraska football -- crunched the numbers and found that the Bendability Index correlated with winning significantly higher than total defense, pass efficiency defense, run defense, third down efficiency defense, and fourth down efficiency defense. The only thing that correlated higher with winning than the Bendability Index was scoring defense, and it did so only very slightly. In the big picture, the Bendability Index may even be a more important statistic than even scoring defense, as was noted by Cold, Hard Football Facts.

So how did my beloved Crimson Tide stack up in terms of Bendability Index in 2006?

Not well at all, unfortunately.

All told, when you run the numbers based on conference play, we finished 11th in the conference in the Bendability Index, ahead of only Mississippi State. At the end of the day, opposing teams generated only 13.39 yards per point scored. By comparison, Florida led the conference, as opposing teams generated over 19 yards per point scored. For another comparison, our 2005 team forced opposing teams to generate 23.75 yards per point scored (I don't have the full SEC data for the 2005, but I'm sure that led the league by a mile).

Certainly that is a sign of a poor play in 2006 by our defense. The loss of the stars of the 2005 defense, being brutally honest, hurt much more than anyone expected. I'm not even sure rival Alabama haters expected the drop-off that we ultimately saw.

But, again, it's not just the defense. It's a measure of overall team efficiency, or in case in 2006, a lack thereof. The Alabama offense turned the ball over entirely too much, and in some cases those turnovers led directly to points. Take Quentin Culbertson's interception of John Parker Wilson returned for a touchdown in the Mississippi State game for example, the Bulldogs put seven points on the board with the offense sipping Gatorade on the bench. But it was also indicative of other problems, particularly on special teams. No one needs to be told special teams were atrocious in 2006. We finished near the bottom in punting, kick-offs were generally short, and we generally allowed big returns. All told, it all combined for opposing teams to have short fields, if not points directly, and that made it much easier to score points by having to move the ball a much shorter distance to get the ball either into the end zone and / or between the uprights.

And, really, that's about all you can say. Unfortunately, it's just another statistic showing just how bad we were in 2006. Hopefully that will change dramatically in 2007 with the new coaching regime.

Either way, the Bendability Index is certainly something you should be keeping your eye on.

Wilson v. The Average

Now that we have relatively advanced statistics regarding the performance of the 2006 SEC pass defenses, we can do a few other things.

For one -- and first and foremost here -- we can compare an individual quarterback's performance versus the average performance of the eight SEC quarterbacks that a particular defense faced in a given year. At bottom, it gives a lot of context to exactly how well a quarterback performed. Raw numbers only tell us, generally speaking, whether or not a player performed well in a game or a season, but comparing their raw numbers to the overall league average tells us not only how well they performed, but how well they performed in comparison to how well other players at the same position performed against those same teams.

So how did John Parker Wilson do as a whole, compared to the overall leave average?

All in all, Wilson's numbers were almost exactly identical to the league average, not really better and not really worse. Being very picky about, Wilson's numbers were generally just a hair below the league average, but by a very small and essentially insignificant amount. All told, Wilson had a higher completion percentage (56.52 for Wilson, to the league average of 56.37), and he also threw for slightly more yards per attempt (7.07 for Wilson, to the league average of 7.00). However, Wilson did have a lower touchdown rate, and a higher interception rate. All told, the combined QB rating of conference opponents against the eight SEC teams we faced in 2006 was 124.3, and Wilson clocked in just a hair below that at 122.4.

In terms of over / under performing the average, Wilson's best day came against LSU in Tiger Stadium. Though LSU held quarterbacks as a whole to under 150 passing yards per game and a completion percentage of lower than 50 per cent, Wilson threw for 291 yards against LSU, and completed 62.8 percent of his passes. Wilson's worst day came, oddly enough, the week before against lowly Mississippi State in Bryant-Denny Stadium. The Bulldogs struggled to defend the pass all year long, as opponents, as a whole, completed almost 60 per cent of their passes for over 220 yards per game, with a 3:2 touchdown to interception ratio. As with most other things that day for the Crimson Tide, it didn't quite go as planned. Wilson completed less than 50 per cent of his passes for under 200 yards total, and threw two interceptions against no touchdowns. Given the succession in games, I suppose an outhouse to penthouse reference would be fitting here.

All in all, though, the results are pretty good. Wilson was a first year starter, suffering from poor coaching and likely poor play-calling. Moreover, the offensive line was poor, and the running game was completely ineffective; arguably our worst since 1993. On top of all of that, Wilson played nearly all year with a badly sprained right ankle (i.e. his plant foot). About the only thing that Wilson had going for him in 2006 was a good receiving corps, and while good (Hall was great), after Brown was injured in the Ole Miss game, we never had another solid receiving threat. All told, while the wide receiver corps was pretty good (due to Hall's performance), it was far from extraordinary. Still, despite all of that going against him, Wilson put up average production numbers in 2007. Not a bad showing at all when you consider the circumstances.

It's just another good sign about the ability and potential of our starting quarterback.

2006 SEC Pass Defense

After posting the analysis of the 2006 Alabama pass defense, I said a post on the 2006 overall SEC pass defense would be forthcoming. Thus...

LSU and Arkansas were two of the top pass defenses in the SEC in 2006. Interestingly enough, though, on the surface you might not expect that they were all that good. Combined, they gave up twice as many touchdowns as they created interceptions, and the Bayou Bengals and the Hogs finished tenth and eleventh in interception rate, respectively. But make no mistake about it, these two pass defenses were damn good. LSU and Arkansas finished first and second in completion percentage, and third and fourth in yards per completion, respectively.

LSU and Arkansas bring an opportunity to make a great point about effective pass defense. In reality, in terms of coverage of pass defense, all that really gets any real attention is touchdown passes and interceptions. That's what Stu Scott blows his load over on SportsCenter, and that is what is replayed a million times on YouTube. But that is far from what effective pass defense is about. As LSU and Arkansas made a great example of in 2006, effective pass defense is mostly about forcing a large number of incomplete passes, and limiting the yardage allowed when a pass is completed. Oh sure, limiting touchdowns and maximizing interceptions are things that you should certainly strive for, but in reality touchdowns and interceptions are a relative rarity, with a touchdown or an interception occurring on only about 8.5% of total pass attempts (8.47% of the time in 2006). Even the SEC leader in interception rate snagged interceptions on only a mere 5.53% of all passes thrown (South Carolina). At bottom, the truth of the matter is that the best pass defense is a pass falling harmlessly to the ground, or if a pass is caught by the offense, tackling the receiver immediately and limiting him to a relatively short gain. If you can consistently do that, almost regardless of what happens with touchdowns and interceptions, you are going to have a good pass defense. Again, LSU and Arkansas proved that point quite nicely with their 2006 performances. On the other hand, if you can't consistently force incomplete passes -- however you go about doing that -- and if you can't limit yardage in the event that there is a completion, you are in major trouble. Even if you snag more interceptions than any other team in the conference, interceptions are such a rarity in terms of overall passes that if you can't force incompletions and limit yardage on the other 95% of throws, you are not going to have a very good pass defense.

South Carolina exemplifies the last point. The Gamecocks in 2006 were great at intercepting the football, as they led the SEC in interception rate. They snagged eleven interceptions on 199 passing attempts, giving them a league-leading 5.53% interception rate. However, they finished seventh in the conference in completion percentage (58.29% of passes against them were completed), and finished tenth in the conference in yards allowed per completion, giving up roughly 13.5 yards per catch. And, of course, the South Carolina pass defense was pretty mediocre, all things considered. Again, it doesn't matter if you are great at intercepting the football, interceptions are such a rare occurrence that you will not be able to build a good pass defense off of just intercepting the football. You must be able to force incompletions and limit the amount of yardage gained when the ball is caught. South Carolina was not able to do that particularly well in 2006, despite their knack for interceptions, and as a result they did not field a particularly good pass defense.

Georgia, too, brings about an opportunity to showcase another point. The Dawgs had a very good pass defense in 2006, no two ways about it. Though they finished sixth in the conference in completion percentage, they were second in yards per completion, second in interception rate, and fourth in touchdown rate. At bottom, this was a good pass defense. But what is the point I'm getting at? The point is that the Dawgs had a very good pass defense, and that was due in large part to a great pass rush from the defensive ends. All told, Quentin Moses and Charles Johnson combined for 15.5 sacks in 2006, and are now playing in the NFL. The point is that good pass defense can come from either a good pass rush from the defensive line, and / or good pass coverage from the secondary. You often times see teams with relatively poor secondaries do quite well against the pass because the rush the passer so well with their front four, and likewise, you often times see relatively poor pass rushing teams do quite well because the secondary covers so well. Georgia proves that point quite nicely. Their secondary -- though stocked with solid players -- perhaps didn't have as good of secondary players as the numbers would suggest, but nevertheless they had a great pass defense because they got such a good, consistent pass rush from their defensive ends.

And that's going to be about it for the lessons, but moving on...

I said earlier that LSU and Arkansas were two of the best pass defenses in the conference, but I didn't say they were the best. Reason being: the Florida Gators. Though the Gators did allow opponents to complete a lot of passes (9th in the conference), they were exceptional in every other category. Granted, they did give up quite a few completions, they generally didn't give up very much yardage. They finished first in the conference in yards per completion, and did so by a very wide margin over second place Georgia. Moreover, despite the fact that teams threw more passes against the Gators than any other SEC team (290 attempts), they allowed the fewest touchdown passes in the SEC, thus leading the SEC in touchdown rate by an incredibly massive margin. Beyond that, the Gators had more than twice as many interceptions as touchdowns, thus leading the conference in interception-to-touchdown ratio, again by a massive margin. What can you say? These guys really had it together in 2006.

The Auburn pass defense was generally regarded as good, but it wasn't anything overly special, and it was a bit intriguing. They didn't give up very many completions (fourth in the conference in completion percentage), but when they did give up completions, they generally went for big yardage (eighth in the conference in yards per reception). Moreover, they didn't intercept the ball particularly well (ninth in the conference in interception rate). However, they did finish third in the conference in touchdown rate, which makes me wonder a bit. It wasn't a particularly great pass defense, so why did they finish so high in that category? A few reasons are possible. One, they may have not given up very many long passes; two, they may have tightened up considerably in the redzone, where teams could have opted more towards the run; or three, they may have just played that much better in the red zone with a smaller space to defend. Either way, it'd be nice to break things down even further to see exactly what was going on there.

Tennessee, as a whole, was pretty middle of the road. They did quite well in a couple of categories, finishing third in completion percentage and fourth in interception rate. But they also didn't do too well in a couple of categories, finishing seventh in yards per completion and ninth in interception rate.

The Ole Miss pass defense was probably a bit better than most gave it credit for being, at least in the eight conference games. They finished fifth in completion percentage, and fifth in yards per completion. They were the definition of average in touchdown rate, finishing sixth in the conference. The problem with the Rebels was interceptions. For whatever reason, Johnny Reb couldn't get an interception if their life depended on it. They racked up a mere three interceptions, and finished dead last in terms of interception rate, with its interception rate a puny 1.72%. Again, though, this was a better pass defense than most gave it credit for. The Rebs only went 2-6 in conference play, but they scared the living hell out of Auburn, Alabama, LSU, and Georgia, and came very close to knocking each member of that group off. Obviously, it wasn't the atrociously anemic offense that kept them in those games, and run defense alone can't do it. The Rebs pass defense, while not particularly good, was pretty decent.

And then we get to the really ugly pass defenses...

Kentucky finished tenth in the league in completion percentage, and as often as they gave up completed passes, those completions nevertheless went for huge yardage. These weren't dink and dunk throws that failed to rack up very much yardage. All told, UK finished dead last in the SEC in yards per completion, giving up 15.66 yards per catch. They were the only SEC team to give up more than 14 yards per reception. Though they intercepted the ball quite well (5th in the SEC in interception rate), they struggled in terms of touchdown rate, finishing 8th. At bottom, this pass defense was just terrible. All year long, teams consistently threw the ball at will with little trouble. How in the world Mike Archer, the defensive coordinator, was hired by NC State, I'll never know.

Mississippi State, too, was equally terrible. They finished eleventh in completion percentage, and eleventh in yards per reception. They did intercept the ball relatively well, sixth in the conference in interception rate, but did give up a lot of touchdown passes, as they finished eleventh in the conference in touchdown rate. At bottom, it was a terrible pass defense. Teams completed a high percentage of passes, and those completions went for a lot of yards. All year long, teams easily moved the ball through the air on the Bulldogs. About the only good game their pass defense had was, unfortunately, against Alabama. What can you say? They went 3-9 for a reason.

And last, and certainly worst, we have the Vanderbilt Commodores. The enlightened 'Dores might win in the game of life, but they didn't win on the field in 2006, and their horrendous pass defense was a large reason why. The 'Dores finished dead last in completion percentage, as teams completed almost 65 per cent of their passes against them. Vanderbilt was the only team to have a completion percentage above 60 per cent. They finished ninth in yards per completion, seventh in interception rate, and dead last yet again in touchdown rate. What can you say? They were just, well, Vanderbilt.

2006 Pythagorean Wins

I've posted previously about Pythagorean Wins in this forum, so perhaps it should be a bit familiar. As a refresher, Pythagorean Wins is a statistical measure of how many wins a team "should" have in a given year based on how many points the team scored and how many points they allowed. The concept was first used in professional baseball, but has since been successfully applied to other sports, and I've applied it to college football as well.

At bottom, the underlying point of Pythagorean Wins in regard to predicting future performance is that teams that overachieve -- i.e. win more games than their Pythagorean projections says they should have -- tend to decline the following year, and teams that underachieve -- i.e. win fewer games than their Pythagorean projections said they should have -- tend to improve the following year. After analyzing SEC football from 1999-2006, it becomes very obvious that Pythagorean Wins is a highly useful indicator of future performance.

Among the major declines that Pythagorean wins predicted included 2000 Alabama, 2004 Ole Miss, 2004 Florida, and 2005 Tennessee. Among the major improvements predicted by Pythagorean wins included 2001 Alabama, 2005 Alabama, 2005 Florida, 2006 Arkansas, and 2006 Tennessee.

Generally speaking, the magic number seems to be one. You come up with a Pythagorean projection based on the eight regular season conference games, and the teams that win a game or more above their Pythagorean projection tend to decline the following year, and teams that win a game or fewer above their Pythagorean projection tend to improve the following year.

Here are the Pythagorean projections for the 2006 season:



Obviously, we had three overachievers -- Florida, Kentucky, and Auburn -- and two underachievers -- South Carolina and Vanderbilt -- in 2006.

The Gators, despite having a projection of only 5.5 conference wins, went 7-1 in conference play, won the SEC, and ultimately became national champions. Auburn was the second most overachieving team, and it shouldn't come as much of a surprise. The Tigers beat LSU on a controversial call, stood up South Carolina on the goal line, beat Florida in a narrow game, needed an interception returned for a touchdown to beat Ole Miss, and had another narrow win over Alabama, despite being annihilated by Georgia and Arkansas. Kentucky, too, becomes suspect with a few close wins mixed in with a few lopsided blowouts.

The underachievers, as largely expected, include South Carolina and Vanderbilt. Both teams played solid football all year, but generally ran out of luck in close contests.

You never know for certain, and there will be some exceptions, but if you are looking for teams to mildly improve in 2006, Vanderbilt and South Carolina would be safe beats, and no one should be surprised if Florida, Auburn, and Kentucky suffer a bit of a decline.

A few more general notes on Pythagorean wins:

• Generally speaking, the underlying rule regarding Pythagorean wins holds true in SEC football. There does seem to be a couple of exceptions, however. The first exception is second year coaches. Even when a team overachieves with a first year coach, they often times nevertheless improve in their second season. It's not a guaranteed thing, but apparently the gains made by continuity and that particular coach's players in place help offset what the Pythagorean projection says you should do. The second exception is with the sisters of the poor of the SEC, mainly Kentucky, Vanderbilt, and Mississippi State. For whatever reason, those teams tend to be able to underachieve one year, and consistently do as bad or worse the following year. I'm not sure why that happens, but it is something that keeps popping up.
  

• If you factor out the last-second fluke touchdown that Auburn scored on Florida (in which had no influence on the outcome of the game), the Pythagorean projections flip a bit, as suddenly Auburn was the most overachieving team and Florida was the second most overachieving team. Some people take out meaningless, fluke points like that because they skew the data, and that wouldn't be a bad idea in this case. But, technically if I did it for that game, I'd have to do it for every other, too, so I decided against that for the sake of simplicity.
  

D.J. Hall: Record Breaker

Anyone who knows anything about football can tell you that D.J. Hall is a great receiver. In 2006, he broke just about every Alabama single season receiving record, and if he stays even remotely healthy in 2007, he'll own essentially every receiving record in Alabama football history.

All of the mainstream statistics for the receiver position, however, are all cumulative, and thus can be inflated by increased opportunities. In short, a receiver can put up big numbers and still not have a particularly good year, with the reason for the big numbers being so many passes were thrown his way. Considering John Parker Wilson set the Alabama record for single season passing attempts in 2007 with 393 attempts, I wondered if that could have possibly been the real underlying reason to Hall's successes in 2007.

So, I applied the usual analysis, I broke down every pass thrown to Hall this season in the eight conference games.

Long story short, the pass attempts didn't inflate Hall's numbers, he was just that good.

All told, in the eight conference games, 78 passes were thrown to D.J. Hall. Of those 78 passes, Hall caught 47 of them, which gives him a catch rate (percentage of passes thrown to a particular receiver that were actually caught by that receiver) of 60.26%, which is a very high number, especially considering the quality of competition faced in SEC play. Hall went up against opposing team's best cornerback, and often times faced double coverage, but yet he still had a Catch Rate of over 60%, a truly impressive number. To give the number some context, generally speaking, the best NFL receivers have a Catch Rate in the sixty-percent range.

And these weren't short throws, either. It wasn't as if Hall was continuously running five yard curl routes, racking up easy receptions. Hall averaged just under 17 yards per catch on those 47 receptions, so his receptions were generally downfield, more difficult, catches.

Hall's catch rate is even more impressive when you put it context with what surrounded him. The quarterback throwing him the ball, John Parker Wilson, was a first year starter, and Wilson was constantly under pressure due to a poor pass blocking offensive line. Moreover, teams keyed on the pass all year long due the complete ineffectiveness of our rushing attack. Beyond that, Hall largely lacked a legitimate receiving threat to compliment his abilities. Keith Brown was hurt in the Ole Miss game, and never really recovered, and no legitimate receiving threat appeared the rest of the year (aside from Stover in the Iron Bowl).

At bottom, opposing teams knew that Hall was all we had, and they put all of their emphasis on stopping him. We threw to Hall 12 times against Tennessee, 17 times against Mississippi State, and 14 times against LSU. Again, opposing defenses knew he was our only real threat down the stretch, but even so they simply could not stop him.

Beyond that, though, the breakdown showed that we didn't exactly throw to Hall in ideal situations. We threw to him a lot on 1st down (36 times), but when we threw to Hall it was almost always in situations where lots of yardage was needed. Granted, that shouldn't be too surprising -- after all, long yardage situations is why you are throwing in the first place -- but it would have been nice if he could have been utilized better. The underlying idea to play-calling is that you should pass when other teams expect a run, and run when the opposing defense expects a pass, and Shula and co. failed miserably at that in 2006 in relation to Hall. Generally speaking, when you have a play-maker wide receiver, you would like to throw him the ball in situations where the opposing defense expects the run, which should free up man-to-man coverage on the outside for your play-maker receiver. We didn't do that at all in 2006. Of the 78 passes to Hall in 2006, only seven of them came with five yards or less needed for a first down. Unfortunately -- another sign of poor play-calling -- we largely threw the ball to Hall when it was obvious we needed to pass.

Beyond all of the foregoing, we must keep in mind one other thing. Hall played almost all of the 2006 season with a serious left shoulder injury, and injury that was severe enough to require off-season shoulder surgery.

Despite all that he had going against him, Hall caught just about everything that was thrown in his general direction. Actually, the play-by-play data at RollTide.com does not list Hall with dropping a single pass that was thrown his way in conference play in 2006, and, moreover, a large chunk of his incompleted passes were desperation throws late in games that were effectively over.

There are no two ways about it. When you break down Hall's performance using advanced statistics, he really was just that good in 2006, and assuming he can stay healthy in 2007, he's going to likely have an even better year, and break essentially every Alabama receiving record.

Hall's big performance is enough to make the case that he is quite possibly the greatest pure wide receiver we've ever had at Alabama.

A Closer Look: 2006 Red Zone Production

In the last post, I took a closer look at our red zone production in 2006, and today we're going to take an even closer look than that.

Thankfully, the stat compilation company used by by RollTide.com this season provided in-depth play-by-play data of each and every snap, so we can analyze the Crimson Tide on a play-by-play basis. Furthermore, I've analyzed only eight games -- the eight conference games -- for the exact same reasons as mentioned earlier in the post on Pythagorean Wins.

So, after breaking down every red zone play in the conference games, what exactly did I find?

In the eight conference games, we ran a grand total of 94 plays in the red zone. And, well, those 94 plays were generally pretty damn ugly. All told, in 94 red zone plays, we accumulated only 166 yards of offense, about 1.7 yards per play. Moreover, we scored only seven touchdowns in these 94 plays, meaning we scored, on average, one touchdown per every 13.4 red zone plays.

John Parker Wilson, despite an otherwise solid year, struggled a good bit in the red zone. Yes, he did throw five red zone touchdown passes, but all told he was only 14-29 in the red zone, completing less than 50% of his passes. Moreover, Wilson also had 10 rushes (either scrambles or sacks) on aborted passing plays. All told, Shula and co. called 39 passing plays in the red zone in 2006, and it resulted in only 14 completions. Putting a percentage to it, only a mere 35.8% of called passing plays actually resulted in a completed pass, and the average completion -- in the relative rarity that it did occur -- went for only about four yards. Unfortunately, the much more often result was a sack, a Wilson scramble, a turnover, or an attempted pass that fell harmlessly to the ground.

Beyond that, I went a bit deeper in breaking down our red zone production into more specific categories. I broke down the 94 plays into four regions: from the 20-16, from the 15-11, from the 10-6, and from the 5 to the goal line. Looking at the numbers in that sense brought about some interesting insights.

From the 5-yard line to the goal line, we were atrocious, as expected, gaining only 15 yards off of 18 plays. All told, when inside the five, we netted on average about two feet per play. And the production from the 20-16 was even worse, as we gained only 32 yards off of 31 plays. Even though it's a higher yard-per-play average than from within the five, it's nevertheless worse because positive gains are capped inside the five yard line by a complete lack of room to move the ball much farther, something that is much less of a problem from the 20-16.

Of course, struggles in both areas had a very negative impact on overall point production. The struggles inside the five yard line turned should-be touchdowns into field goals, and the struggles to move ball from the 20-16 inside the 15 not only hurt point production by stopping drives that could have potentially produced touchdowns, but also by making the field goals that had to be kicked more difficult because of increased distance, which is course drives down the likelihood that those kicks will be successful. And the latter impact is far from insignificant. Given the relative inconsistency of college kickers, a field goal five to ten yards further away probably has anywhere from a 10%-25% smaller chance of being good than from one five or ten yards closer to the goal line.

Oddly enough, though, between the 15-6, we did relatively well, and much better than we did from the 20-16 or the 5 to the goal line. In that area, for whatever reason, we averaged around three yards per play, which, while not great, is triple what we averaged from the 20-16 and the 5 to the goal line. I honestly have no clue as to why that was. Perhaps it was just a statistical oddity that would have worked itself out had we seen a larger sample size, but perhaps there was a legitimate reason for the differences in levels of production. If it is the latter, though, I can't think of why that would be the case.

But what about offensive balance?

All we heard all year long, and even up until today, was that one of our main problems in the red zone was that we were not balanced, that we ran the ball entirely too much and got conservative when it counted; opposing defensive coordinators knew what we were doing, and this allowed opponents to effectively stop it. So how does that argument play out?

Once you run the numbers, that argument doesn't seem to have that much validity, despite spirited assertions to the contrary. All told, from the 20-6, we had 71 plays. Of those 71 plays, 39 were runs (54.9%), and 32 were passes (45.1%). And, of course, that 55-45 split is generally the definition of offensive balance in college football. Even when you break it down by individual games, we were still pretty balanced. We weren't too balanced in the Vanderbilt game (ran mostly), the Arkansas game (ran mostly, especially late in the fourth quarter), and the Florida game (passed mostly), but for the other five SEC games, we were really well balanced in terms of play-calling from the 20 to the 6.

At bottom, despite what some people will strongly assert to the contrary, conservatism wasn't the reason we struggled in the red zone. We were atrocious, for the most part, regardless what we did, whether it be run or pass.

From the five yard line to the goal line, it was a bit of a different story. We had 18 plays from inside our opponent's five yard line, 13 runs and 5 passes. Obviously, the play-calling was heavily swayed in favor of running the football, but before you criticize Shula, you have to keep in mind that the overwhelming majority of plays inside the five yard line should be running plays. Throwing the football that deep in your opponent's territory is very difficult, and the interceptions that can occur generally yield much higher return yardage for the opposing team than a normal interception. All told, running the football generally yields small, consistent gains, and that is what you need with such a short distance to go.

The problem wasn't, seemingly, the fact that we ran the ball so much from inside the five yard line, it was who we ran it with that was the problem. In conference play, Tim Castille -- out of the dread jumbo package -- got eight carries inside the five yard line (Castille carried the ball on almost 45% of plays inside the five), and those eight carries netted a grand total of five yards. All told, Castille averaged about 20 inches per carry inside the five yard line. In general, with Castille -- a fullback -- plays were very slow to develop, and it gave opposing defenses a lot of time to shoot the gaps and stop the play for little or no gain. At the end of the day, hindsight 20/20, a quicker back like Johns or Darby would have likely been the better solution.

The bottom line, the jumbo package should have be re-titled the junk package.

So what all should we take from the analysis? I'd say two things.

One, the play-calling -- or, more precisely, I should say the supposed overly conservative play-calling -- was not, apparently, the problem with our lack of red zone production in 2006. One way or the other, regardless of what we tried in the red zone, it just really didn't work. Running plays were generally stuffed at the line for little or no gain, and passing plays more often than not resulted in incompletions, interceptions, fumbles, or sacks. Our lack of red zone productivity would most certainly not have been improved simply by throwing more passes and opening the offense up more.

Two, however bad you thought our red zone production was in 2006, the statistics show that it was probably much worse than you imagined.

At bottom, after really going in-depth on overall red zone production in 2006, you just appreciate how horrendous it really was. Odds are, we could perform poorly in the red zone again in 2007, and still have red zone production significantly better than in 2006.

Heaven Help Us: Red Zone Breakdown

The Crimson Tide was atrocious in the red zone in 2006, and no one needs any in-depth analysis to figure that one out. But exactly how bad were we, and how much did it hurt us?

Answering the first question, it was as bad as you think, and then probably worse.

After running the numbers, we made 53 visits to the red zone in 2006. Of those 53 visits, we scored on just over 75% of them (40), which is a good -- but certainly not great -- number. The major problem, of course, was being able to put the ball into the end zone. All told, we only scored 20 touchdowns in 53 trips to the red zone, thus putting it in the end zone on only 37.74% of red zone trips.

It gets worse.

When we look at the eight conference games the Tide played (which, of course, we went 2-6), we made 28 trips to the red zone. Of the 28 times that we got into the red zone in SEC play in 2006, we managed a mere 6 touchdowns. 6 touchdowns in 8 conference games. Long story short, we scored a touchdown on only about 20% of our red zone trips in conference play.

Now let's compare that to LSU, who led the conference in both total offense and scoring offense. Believe it or not, LSU had essentially the same number of red zone trips in conference play that we did (LSU had 30 red zone trips, we had 28). But, of course, they did much more with theirs. All told, LSU scored 22 touchdowns in their 30 trips, and, well, that was a big part of the reason why they led the conference in point production.

So exactly how much did the lack of productivity in the red zone hurt us in terms of wins and losses? Quite a bit, it would seem. With even half-way decent red zone production (say, just to quantify that, 14 touchdowns on 28 trips), our final win-loss record would have been drastically different, and the margin of victory in the games we did win would have gone up as well.

Against Hawai'i, despite us leading all night, it turned out to be a close game, with victory ensured only when a time-expiring Colt Brennan heave to the end zone was intercepted by Lionel Mitchell. We went to the red zone five times that night, and emerged with only one touchdown. With decent red zone production, that game never comes down to a last-second heave.

Against Vanderbilt, the following week, we narrowly edged out a 13-10 victory, at home, over the eventual 4-8 Commodores after Leigh Tiffin booted a 47-yard field goal late in the fourth quarter. During that game, we went to the red zone four times, scored no touchdowns, and came away with only six points. Again, with decent red zone production, that game is never that close; we would have by somewhere in the neighborhood of 14-20 points.

Against Arkansas, two weeks later, five trips to the red zone netted only one touchdown and ten points. Even with slightly better than atrocious red zone production, much less decent production, Alabama wins that game in Fayetteville. Instead, horrendous red zone production turns a sure-win into a one-point loss.

The following week against Florida, two red zone trips netted six points and no touchdowns. Decent red zone production might not have turned this game into a win, but it would have made it very close. Most people forget that this was a 14-13 game with under seven minutes to go in the fourth -- and Alabama having the ball. With decent red zone production, we likely have a 17-14 lead at that juncture, and instead of throwing the football -- which led to a Reggie Nelson interception returned for a touchdown that ultimately doomed us -- we are running out the clock. Again, Florida may still have won, but it would have been a much closer game.

Against Ole Miss -- big shock forthcoming -- we actually had solid red zone production. I've written before that it was probably our best offensive performance, and rightfully so. Four trips to the red zone resulted in 20 points (two touchdowns and two successful field goals). And -- big shock again -- we won this one, though one of the poorest defensive efforts of the year turned it into a very close game.

The following week in Knoxville, poor red zone production was largely responsible for a loss to the hated Volunteers. We had three trips to the red zone, and could muster only one touchdown. With just one more touchdown on the other two red zone trips (again, my definition of half-way decent red zone production), we win that game. Instead of leading 13-9 late, we would have been leading 17-9, and Tennessee would have not only needed to score a touchdown, but successfully complete the two-point conversion just to force overtime.

Against Mississippi State, it was more of the same. Four trips to the red zone netted zero touchdowns and only nine points. In a game ultimately decided by eight points, decent red zone production could have gotten us a win.

One week later against LSU, we played the Bayou Bengals surprisingly tough, but again... red zone woes. We made three trips to the red zone, which netted only ten points. Although this is most likely still a win for LSU with decent red zone production, it would have went right down to the wire in a close contest, particularly if Jamie Christensen could have connected on a first quarter field goal.

Against Auburn, in the Iron Bowl, red zone woes played another big role in our loss. Three red zone trips resulted in only one touchdown, and nine total points. Again, with decent red zone production, we may win that one.

At bottom, red zone woes really killed our 2006 team. It wasn't just that our problems in the red zone hurt us some, they have a drastic effect on our overall win-loss record. We finished the regular season at 6-6, but being quite objective about it, and decent red zone production would have probably given us an 8-4 or 9-3 regular season record, with those losses being very close losses at that. That would have put us, likely, in the Peach Bowl, and Mike Shula would still be the head coach of the Crimson Tide.

The good news though, I suppose, is that with nine starters returning on the offense for 2007, this group should be able to finally break through and translate all of that potential into points.

2006 Season: How Johns Was Under-Utilized

As mentioned in the previous post, it was much the debate last Fall over whether or not Jimmy Johns was being utilized correctly. After charting every carry Johns had last season over the course of our ten big games, we can take a much closer look at that.

Against Hawai'i, Johns had seven carries in the first half. All seven carries were successes when judged by the Running Back Success Rate, and the seven carries resulted in a grand total of 56 yards, thus averaging eight yards per carry. Moreover, five of the runs either went for a first down, or put the Tide within one yard of a first-down. Yet, for some reason, after picking up nine yards on a second and ten in the second quarter, Johns didn't see another carry until the fourth quarter, and ultimately that carry would be his last of the evening.

The following week against Vanderbilt, it was more of the same. Johns had five carries in the first quarter, three of which were successes, and they ground out 25 yards of rushing. After the initial success in the first quarter, though, Johns didn't see another carry until mid-way through the third quarter, which resulted in a fumble (backup-itis?). All told, after his hot first quarter start, Johns saw two carries the rest of the day.

Against Arkansas, it was another solid start, with four carries for 18 yards in the first half. Then, inexplicably, he dropped off the map. After that, Johns only saw one carry (a draw play on a 3rd and 26) before finally being inserted at the end of the fourth for some token carries to run out the clock, and presumably set up a successful Leigh Tiffin field goal.

Against Ole Miss, more of the same. Johns started out with 29 yards off of just four carries, three of which were successes, in the first quarter. From that hot start, he only saw one carry the rest of the day.

Against Tennessee, Johns was hot early yet again. He had three carries in the first half for 38 yards, including a 26 yard scamper. All three carries were successes, and either picked up first downs for the Tide or put them within two yards of a first down. Yet, after the hot first half start, Johns didn't see another carry for the rest of the day.

Same thing again when we played LSU. Johns started out with four carries for 25 yards in the first half, two of which were successes. Yet, after another good start in the first half, Johns never saw a carry in the second half.

Finally, it all culminated in the Independence Bowl, where Johns again looked good early. In the first half, Johns had 38 yards off of just five carries, four of which were successes. Yet, after the initial first half success, Johns only saw one carry in the entire second half, and if my memory is correct, that was an attempt at a bit of a gimmick play.

The story just keeps repeating itself. Johns would have a lot of success on carries in the first half, and then Shula would rarely, if ever, call his number in the second half.

In the seven games I just mentioned, Johns had 33 first half carries for 229 yards. On average alone, that's 6.93 yards per carry, but moreover, an amazing 24 of his 33 carries were successes, giving him a success rate of 72.72%. Long story short, Johns' productivity on the limited first half carries he had was simply incredible.

And then Shula, for whatever reason, decided Johns shouldn't get the time of day in the second half. All told, despite the great production in the first half from Johns, Shula rewarded him by giving him less than 10 second half carries over a seven game stretch. Even with a stagnant offense and an even more stagnant running game, Shula chose to keep what was statistically by-far his best back on the sideline.

Small wonder Shula is no longer at the Capstone, and an even smaller wonder that he is not calling plays with his new position with the Jacksonville Jaguars.

Analyzing Jimmy Johns

Jimmy Johns and his performance in 2006 was very much the hot topic of discussion last Fall. For months, almost from the first snap of the Hawai'i game to the end of the Independence Bowl, Darby v. Johns debates filled Alabama message boards and other mediums of football discussion. Of course, the performance of Johns is not purely one of past relevance for the Crimson Tide. With Darby gone, Johns has the most experience of any returning back, and his performance will likely be a key determinant of the success in our rushing attack in 2007.

Thus, I decided to break down John's performance in the 2006 season.

Just scratching the surface, using the standard statistics, Johns logged 66 carries for 293 yards, and thus averaged 4.4 yards per carry. He had one touchdown, and his longest run from scrimmage was 26 yards. Of course, the purpose of this blog is to go much deeper than anything you will find in a traditional stat-line, so let's start delving further right now.

For starters, I charted every carry that Johns had in ten games this season. The eight games included the standard eight conference games, the season opener against 11-2 Hawai'i, and the Independence Bowl vs. Oklahoma State. I chose to ignore the results from the Duke game and the Florida International game, mainly because both were such terrible teams (0-12 each), and many of the carries by Johns in both games came from the quarterback position where he was running Shula's variant of Meyer's Tim Tebow package, which does us little or no good in evaluating Johns' effectiveness as a tailback.

All told, in those ten games, Johns accumulated 51 carries for 274 yards. However, I decided to omit three of those carries for a variety of reasons (one was a halfback pass gone bad, another was a draw play on third and 25+, and the other was an intentional carry solely to move the ball to the middle of the field at the end of a half). Once adjusted for those carries, Johns carried the ball 48 times for 267 yards, thus averaging 5.56 yards per carry.

Impressive as that yard-per carry number may be, Johns' performance looks better the deeper we go.

Football Outsiders came up with Running Back Success Rate in order to take down and distance into account in whether or not a particular run is successful. Long story short, a run is considered successful if it generated 40% of yards needed for a first down on first down, 60% of yards needed on second down, and 100% of yards needed on third and fourth downs. Obviously, the higher the success rate for a tailback, the more he consistently picks up the needed yardage, and the lower the success rate for a tailback, he's less consistent in picking up the needed yardage. In the NFL, a success rate of 50% or higher is considered very good, while a success rate of 40% or lower is considered very poor. So how did Johns stack up? Of his 48 runs, 28 were successes, which means he had a whopping 58.33% success rate. Though I would imagine success rates for good backs in college football tend to be a bit higher than in professional football (due to greater disparities in talent between good and bad players), a success rate of almost 60% would be considered very high. Any time a player, at any level, can create a successful play almost 60% of the time he touches the ball, it's speaks highly of that particular player, especially when he does so with an offensive line as poor as the line we had in 2006.

Beyond that, Johns did an exceptional job of limiting negative plays. Football Outsiders, again, came up with a statistic to measure negative plays by tailbacks, called the Stuffed Rate. Technically, a back is "stuffed" if he gains zero or negative yards on first down, or if he gains less than one-fourth of the yards needed for another first down on second and third down. Using that metric, Johns was again impressive. Of his 48 carries, he was stuffed a mere 6 times, for a Stuffed Rate of only 12.5 percent. Again, a very impressive statistic that speaks volumes of Johns performance. On the whole, studies have indicated that backs that consistently gain positive yardage have a higher value than backs who are more boom and bust -- those who have big gains sprinkled in with several no gains and negative plays, and that seems to fit Johns to a T. All told, he consistently gained positive yardage at a very high rate all season.

Another thing to consider, largely on those lines, is how a few long gains can impact a back's average. In reality, a couple of very long gains in a season when a back has a relatively few amount of carries can change a tailback's yard-per-carry average by two yards or more (Tide fans may well remember when Santonio Beard averaged over 7 yards per carry in 2001 when that very thing happened). As a result, a few big gains skew the data to make it seem like a particular back consistently gained more yardage, when in reality he wasn't as good, but had a couple of big runs along the way (Again, Beard, 2001). You can get around that, statistically speaking, by capping the runs of more than 10 yards and just counting them as a 10-yard run. For example, a 75-yard run goes down statistically as a 10-yard run. So what effect does that have on Johns' performance? Again, Johns performs well here. Johns had 6 carries of 11 yards or more in 2006, and once you cap those runs at 10 yards, Johns ran for 224 yards on 48 carries, which still yields a yard-per-carry average of 4.66. Again, even when you cap the big runs to keep them from skewing the average, Johns still shows through as a consistent yard-gainer almost every time he touched the football.

Let's go one step further by categorizing Johns' carries by yards gain on each individual carry:

Negative to 2 yards: 14 times (29.17%)
3-5 yards: 15 times (31.25%)
6-10 yards: 13 times (27.08%)
11+ yards: 6 times (12.50%)

Again, Johns looks very impressive. Over 70% of his carries went for three yards or more, and right at 40% of his carries yielded six yards or more. Again, the basis is the same: Johns consistently gained good chunks of positive yards each time he touched the football with relatively few negative plays mixed in.

I, for one, have been a bit harsh on Johns and his performance in 2006. However, perhaps next time I should keep my mouth shut before I do a more thorough analysis. After analyzing his season much closer, Johns looked very good in 2006. Certainly, he needs to improve greatly on the fundamentals of being a tailback, but even with poor fundamentals, he still had a very solid and productive season in 2006. Much of that, of course, can be chalked up to Johns' amazing physical abilities, something everyone has gushed about since he stepped onto campus over two years ago. Without doubt, Johns has a lot of work to do off the football field, but on the football field he looked very good in 2006.

Judging by his performance in 2006, mixed with his amazing natural abilities, if Johns can get it together off-the-field and with his fundamentals, he seems to have the potential to be a truly great back.

Offensive Unbalance

One of the most insightful indicators of offensive balance is the play-calling on first down. With first down and ten, head coaches and offensive coordinators can literally do whatever they want without being overly influenced by more extreme down and distance situations. Quite obviously, given freedom to generally do as they please, it becomes of the utmost importance for coaches to find offensive balance on first and ten. By looking at run / pass ratios on first downs, we can see which teams were the most balanced, and moreover, by looking at which plays teams called in the first half on first downs (where they are less influenced by trying to play catch-up late in games with lots of passes, or trying to run out the clock with several runs), we can get an even purer look at offensive balance.

So how did the Crimson Tide stack up in 2006 in terms of offensive balance in first downs?

Unfortunately, not too well. For the entire season, we were very balanced offensively, with 112 running plays and 111 passing plays on 223 first down plays. Statistically speaking, that's 50.2% run, and 49.8% pass. Balanced, anyone?

However, looking more closely at the individual games, the Tide was, generally speaking, unbalanced offensively, and very unbalanced on first downs in the first half. At times, we ran the ball entirely too much, and at others we threw the ball entirely too much; rare was the occasion in which we achieved any real balance.

Against Vanderbilt, we ran 59.26 percent of the time on first down, and 75% of the time on first down in the first half. Two weeks later against Arkansas in Fayetteville, we ran the ball on 70.37% of first downs, and 64.29% of first downs in the first half. Although the total number of runs were inflated by an attempt to grind out the clock late in the game with hand-offs to Ken Darby, the first half numbers show that we were running the ball at almost a 2-1 ratio even then. It was the same story when we played Tennessee in Knoxville, when we ran the ball 72.73% of the time on first down in the first half.

On the other hand, on a couple of occasions we threw the ball entirely too much. Against Florida, on 29 first down plays, we threw the ball 21 times, i.e. 72.41% of the time. More amazingly, in the first half we had 13 first down plays, and threw the ball on 12 of them. Small wonder we managed a net of only six offensive points in that game. Later in the year, against LSU in Baton Rouge, we threw the ball 73.68% of the time on first down. Of the 19 first down plays that occurred in the first half, we ran the ball only 5 times.

Honestly, the only SEC game that we found a good balance on first down was the Ole Miss game. There, in the first half, we had twelve first down plays, and ran the ball six times and threw the ball six times. As an entire game, we ran the ball a good bit more on first down than we threw it, but that was because Ken Darby took over in the second half and found great success. Certainly no blame can be found there for continuing to give the ball to Darby.

At bottom, the point production was atrocious in 2006, and that was due to a large variety of reasons. However, breaking down the run/pass ratios, we can see one of the major problems of the 2006 offense was a complete lack of offensive balance in conference games. Long story short, the offense was just so predictable in terms of play-calling, and that, too, likely had a very negative impact on overall production. It wasn't the only problem, of course, but it was one problem.

Hopefully the new Saban / Applewhite offense will be able to find the play-calling balance that the Shula offense couldn't find in 2006.