A historical and analytical look at UF's Red Zone Offense, Tim Tebow, Play Calling

[Matthanuf06]

Good evening Gators. I decided to actually do some research and analyze our red zone tendencies instead of continue on with the bickering this board has been full of the few days

Of course much more analysis will need to be done to get a strong picture, but I wanted to see the changes over time. No matter what conclusion you draw, and I hope we can have a healthy debate, please remember that this is only a piece of the pie and there are many different ways to look at it. Feel free to explain why this is worthless or why this is important, what you feel is interesting, what you feel is irrelevant, and everything in between.

Process
My goal here was to look at how we call "Tebow Centric" plays (I'll define later) in the Red Zone and compare it to our TD% over 2007-2009. I wanted to see if there was a change in 2009. I wanted to see if there was a relationship between these "Tebow" plays and how often we score touchdowns.

A "Tebow Centric" play (or TT for short) in the red zone, IMO, is a Tebow run on 1st down, a Tebow run on 2nd down, a Tebow run on 3rd down, or a Tebow pass on 3rd and 4 or more. This is of course open to interpretation, but I don't feel a pass on 1st, 2nd, or 3rd and short is truly relying on Tebow. Even if you disagree with this definition don't worry about it, the point is to measure the change of the play calling over time. I only used games vs SEC opp, Miami, FSU, Michigan, and OU.

So I took these TT Plays/All RZ Plays as well as TD/RZ Opportunities to get my percentages. I'm all for people checking my work as I have no doubt I missed some here and there, although I don't think it would change the findings. Also I used some judgment. For example if it was 1st down on the 22, we threw it for 8 yards, and had 2nd and 2 on the 14 I'd count that as a Red Zone opportunity. However if we didn't hit the RZ until 4th down, or say on a 3rd and long situation I didn't count it. Of course I didn't use Brantley possessions in this analysis.


Conclusion-For those who don't need the details
There is most definitely a change in play calling. This is without a doubt. The % of "Tebow" plays is the highest it has ever been, and the closest it has been to this ratio was early in 2007, which is HEAVILY influenced by the Auburn loss. The data also shows that a higher % of "Tebow" plays should result in a lower TD%. As we know it does. However, the data also shows there is no way it should be to this extreme. Therefore it is safe to say there is a luck/execution component. Even if play calling stays the same we should see more touchdowns. If play calling changes we should see a dramatic increase as our luck is due to change.

Interesting Tidbit that you'll see later or you can read now
The data only shows two outliers in 2007 and 2008. Guess what games they were? Auburn 2007 and Ole Miss 2008.



Data
I like graphs so hopefully you do to. Here is the core data and the percentages. Obviously the percentages can be a bit skewed depending on the number of opportunities (eg 100% TD% in 1 of 1 opportunities). It is 2007 to 2009, left to right






Firstly, the first thing that stuck out at me was Auburn '07 and Ole Miss '08. Our reliance on Tebow (as a % of our RZ plays) was extreme. Secondly the only time the % lines "cross" is beginning in 2009. Honestly I wouldn't take a lot of these game by game numbers, but that is the data.


Here is the same data, sorted highest to lowest.






Now just for a reference point lets look at our PPG. Here are the raw numbers, as well as a 4 game rolling average to smooth things out.



Two things stick out. Firstly we are at our lowest 4 game average in the Tebow era. Not good. Secondly we are very streaky. A 4 game average should be enough to balance out the highs and lows to produce something more stable, yet we still see our points rise in fall in streaks.


Now onto the most important graph. It is an attempt to smooth out the TT plays and the TD% graphs so we can identify trends. It is somewhat like a rolling average, but essentially I just got the % of the previous 4 games. This is what it comes out like.

If you need to see anything in this post this is it.





Firstly, as we all have expected, our TD% has absolutely crumbled. Our 4 game average has consistently been in the 65-80% range, yet we've dropped to extreme lows.

Secondly our 4 game TT% average is at its highest levels ever. The spike we see in 2007 is due to the Auburn game (which is the highest game on date). If we remove that (or just go a few downs on the graph) you'll see the TT% is consistently in the 35%-45% range. It is now far higher approaching 60%. As you can see as the TT% increased, the TD% decreased. Now is that the only reason? No of course not. This is just information that many of us would like to see.



Now lets look at the correlations. Not super high, but they exist.





Okay now lets take a peak at how we've performed in wins vs losses.



Our "TT%" is far higher in games we lose. My guess is that is a combination of things. Firstly it is probably an over reliance. Secondly in blowouts we probably take it a bit easier on our best player. However our TD% is not much lower in the losses. This number is a bit skewed due to the Michigan game (5 of 6) which heavily influences this number.


Conclusion
As I stated earlier this bout of poor red zone performance is both luck/execution and play calling. I ran a regression (R-Sq in the mid 60s) and that says that even with our change in play calling (increased reliance on Tebow) that we should not be performing this poorly. This model says we should be scoring at around a 63% clip in the redzone given this increased reliance on Tebow. We are currently scoring in the 25%-40% range. That difference is the luck/execution component. However the 63% by itself, would still be the lowest in the TT era. Typically we should score TDs at a 75-80% clip. That difference is play calling.

Of course, as I've stated previously, there are tons more variables and this is only a glimpse into what is happening. Hopefully we can discuss all of the other possibilities as well as this study in this thread. And before you say it, yes I trust the coaches. I think Meyer is the best coach in football and will be the best coach of all time. That doesn't mean he is perfect though. Go Gators.

[michplaced_gator]

Holy Crap!

...now do an analytical look at Tim's throwing motion before and after Scot Loeffler and we might have what our issue is.

[Lawdog88]

Excellent work Matt, and I appreciate the effort.

One question though (from my simple mind), in your final conclusion, how is it that you attribute the difference in the "should be scoring" at the 63% rate to the "we traditionally scored" at 75-80% rate to play calling? Is it because that is the last variable not accounted for by the other factors / conditions?

Seems to me the flip side of failed execution is that the Defense of the other team has something to do with failed plays that is a different variable / factor than our own "failed execution" or bad play calling.

Wahdda I know from statistics though.

[Matthanuf06]

Originally Posted by Lawdog88

Excellent work Matt, and I appreciate the effort.

One question though (from my simple mind), in your final conclusion, how is it that you attribute the difference in the "should be scoring" at the 63% rate to the "we traditionally scored" at 75-80% rate to play calling? Is it because that is the last variable not accounted for by the other factors / conditions?

Seems to me the flip side of failed execution is that the Defense of the other team has something to do with failed plays that is a different variable / factor than our own "failed execution" or bad play calling.

Wahdda I know from statistics though.

I ran a regression of our 4 game 'averages' and utilized that formula to determine what clip we should be scoring at. Of course there are many other variables and it certainly isn't a linear function. However it is a good starting point. The "traditionally scoring" number is just me eyeballing where the 4 year 'average' number usually is at. If you look at the 4th chart you'll see the TD% range is in the 65-80 from 2007 to the start of 2009 and it averages out closer to 75%.

Another way of looking at it is if you look at the graph the TT% right now (far right of the graph) is pretty much the same as the number to the far left of the graph (first point in 2007). The TD% on the far left is in the mid 60's.

What is interesting is when the TT% drops there is a corresponding boost in TD%. When it increases the TD% drops...

[Bongogator]

OMG! Thank you so much. The 4 fumbles and int's don't help the #'s. So, it's not that bad.

[AnthroMatt]

Nicely done. Has someone emailed this link to the coaches?

[Matthanuf06]

Originally Posted by Bongogator

OMG! Thank you so much. The 4 fumbles and int's don't help the #'s. So, it's not that bad.

The TOs is precisely the "luck" component that I was referring to. Unfortunately there looks to be play calling issues as well.

[HotlantaGator]

Great job Matthanuf06...

that last graph really illustrates a couple points:

1) the PPG UF was scoring in the back half of 2008 was probably a high point, and we just can't expect to score 40-50 points/game in SEC play....

2) On the flipside, the relationship b/w the increase in "Tebow plays" and scoring fewer PPG seems to be pretty strong...

I wonder, Tebow is such a special and great player, that when we lost Harvin it would have been easy for the coaches to lapse into relying on Tim to an excessive amount, but the idea of relying primarily on one player runs counter to the entire idea of their spread offense, you need to both ride your best players but you have to spread the ball around and you need to get the ball to your weapons in space.

Letting Tebow keep the ball so much, b/c he is the QB behind the C in the middle of the field, just isn't the spread offense...

[BJ1888]

First, this is a hell of an effort. But I'm going to have to question some of the findings...

1) What's the policy argument for throwing out random games? Citadel and CSU should probably be thrown out, but Hawaii and Troy are 1A teams. This is important later...
2) Rolling averages aren't that helpful when they include games that happened 11 months prior. Does our scoring against FSU in Nov '08 have any bearing on our play against UT this season?
3) The scoring of a team, "in general," will increase as the season goes along (when you throw out the cupcakes), and then fall off when CCG and Bowl season comes around (better quality opponents). Is it weird to see our scoring drop when you substitute 'Bama and OU for UGA and Vandy? I don't think so.
4) Some of those 4-game averages are flat wrong. Our average went UP this week. We scored 6 more points vs MissSt than against UT, so the average should've gone up by 1.5 points. Last year, we scored more against USC than LSU, yet our average goes down on your chart...
5) I think your assessment of the plays in the RZ is a little off. Arbitrarily deciding whether to count a play makes the data for overall RZ success % totally bogus.
6)Here's the data for the UT game this year (I will not count FG attempts as "plays" because we weren't trying to score a TD). "Tebow Plays" in BOLD. I count 8 of 15 plays as "Tebow Plays" (including a 3rd-1 run and a 1yd TD run)

F 1-10 T19 [shotgun/1 back], Demps, J. rush up middle for 3 yards to the UT16 (Rico McCoy).
F 2-7 T16 [shotgun/0 backs], Tebow, T. sacked for loss of 14 yards to the UT30 (Ben Martin), PENALTY UF intentional grounding --- called pass doesn't count...

F 1-10 T16 [shotgun/1 back], Tebow, T. rush over right guard for 4 yards to the UT12 (Dan Williams).
F 2-6 T12 [shotgun/1 back], Tebow, T. rush over right end for 5 yards to the UT7 (Eric Berry). scramble from pass formation. --- called pass doesn't count...
F 3-1 T07 [shotgun/1 back], Tebow, T. rush over left guard for 6 yards to the UT1, 1ST DOWN UF (Janzen Jackson).
F 1-G T01 [shotgun/2 backs], Tebow, T. rush over left guard for 1 yard to the UT0, TOUCHDOWN

F 1-10 T17 Tebow, T. rush over right tackle for 6 yards to the UT11 (Rico McCoy).
F 2-4 T11 [shotgun/0 backs], Tebow, T. right flat pass incomplete to Cooper, R. (Ben Martin).
F 3-4 T11 [shotgun/0 backs], Tebow, T. crossing pass incomplete to James, B. (Janzen Jackson).

F 1-10 T19 [shotgun/0 backs], Tebow, T. slant pass incomplete to Cooper, R..
F 2-10 T19 [shotgun/2 backs], Tebow, T. rush over left end for 7 yards to the UT12 (Brent Vinson).
F 3-3 T12 [shotgun/1 back], Tebow, T. rush up middle for 5 yards to the UT7,
1ST DOWN UF (Eric Berry).
F 1-G T07 [shotgun/1 back], Demps, J. rush over right guard for 7 yards to the
UT0, TOUCHDOWN

F 1-10 T17 [shotgun/2 backs], Tebow, T. rush over right end for 4 yards to the UT13, out-of-bounds (Dennis Rogan).
F 2-6 T13 [shotgun/2 backs], Tebow, T. rush over right end for 11 yards to the
UT2, fumble forced by Montori Hughes, fumble by Tebow, T. recovered by UT

so 53% Tebow and 40% TD rate (maybe total different w/o Tebow fumble)

7) Kentucky this year:

UK RZ Drives:
2-4 FG
2-3 TD (TD came on 3rd-1 from 3)
2-2 TD (TD on 1st-G from 1)
2-4 FG (last play was Tebow's concussion)
Total: 61% 50% TDs

8) LSU this year (concussed Tebow):

LSU RZ Drives:
2-3 FG
1-4 MISSED FG
2-3 FG (last play was a Tebow run on 3rd-11)
Total: 50% 0% TDs

9) MSU this year:

MSU RZ Drives:
1-1 MISSED FG (3rd-5 from 20 incomplete)
3-3 FG (7yd run, then stuffed twice from the 1)
2-3 FG (sack, OOB for loss (scramble?), throw away)
1-3 (Demps, sack, INT for TD)
2-3 FG (incomplete pass, offside penalty, 4yd run, stuffed on 3rd-1 from 9)
3-4 TD (start 3rd-4 from 16; short pass, 4th-1 run, run, Rainey TD)
Total: 70% 17% or 68% 20%

Tebow was awful here. Stuffed 3 of 4 times on 1yd-to-gain tries, another loss on a run, 1-5 with a Pick-6... TV crew floated the idea that Mullen knows the short-yardage blocking schemes we run. Makes sense because we could not pound the ball on the goal-line at all.

[insuragator]

My alcohol-trimmed-brain cells are in total cluster-*&(^ mode right now.

[Matthanuf06]

Where did you get your info? I'm sure I missed some esp when it says a run when it was a pass play or vice versa. In fact you missed one yourself in the UT game. 1-10 on the third drive.

Re: Why I tossed games out. I just didn't feel like it was apples to apples. Probably my own personal bias though. I just don't see the team behaving the same way they would against CSU than they would vs LSU. But perhaps a team like Troy should make the caught

Re: Rolling numbers. You are right in one sense. It isn't very helpful to look at a rolling number that has two 2008 and two 2009 data points when trying to see if there is a trend. My main goal wasn't really trend hunting though, although that was something that did pop up. I was more interested in the relationship between the two. I wanted to get the sample size high enough for each data point to get rid of some luck or just a low sample in a particular game. For example a game just has 1 red zone opportunity. Or a few games have an inordinate amount of starting drives deep inside the 5. I thought 4 games would be a decent amount to balance that out.

Re: The scoring. I probably shouldn't have streaky, a better term is cyclical. I just wasn't expecting such a pattern with a 4 game average. I think you are overstating the defenses in these 4 game samples. For example the average of Miami, UT, Miss, Arky last year was 31 points....the average of LSU, UK, Vandy, and UGA was 52.5 points. Was the defenses in defense that large?

Re: The points graph. Thanks for the catch. My numbers were off (I was using 3 year averages for some reason). However now the correlations are even stronger which is a good thing.

Re: Arbitrary deciding what a RZ opportunity is. I don't think this is a big deal. The only opportunities I didn't count were field goals if they entered the RZ on 4th down and 3rd and long if they were on the 19 or 20. To be honest, if I added those numbers that would just increase the strength of this study because those plays were typically Tebow passes that didn't result in a touchdown. I just didn't think it was fair to pin a lack of a TD on the RZ offense in such a scenario.

Gotta update the graphs though. Thanks

[Matthanuf06]

Ok I updated the graphs. On the points graph you can now see the uptick that should have been there. The correlation between TT centered plays and points is much stronger now though

[TrumanCoyote]

I'm not sure how useful points per game is in this discussion. Lots of those points have been scored by the defense and special teams. Perhaps points per possession would have been more useful if you are trying to measure the efficiency of the offense.

[BJ1888]

I go off the box-scores and season reviews on gatorzone. totally missed that one when I was doing the "bolding." so it should've been 9-16 (56%)

I think that the data shows that when Tebow plays well, and is asked to do what we know he's best at, we score more. If he picks up the short yardage, and makes completions to move the chains, we do better. When he gets stuffed on short yardage (MSU), or doesn't complete the passes we ask him to (also MSU), bad things happen.

I'm not really comfortable reading a lot into the LSU performance, just based on the health issue.

Re: The scoring. I probably shouldn't have streaky, a better term is cyclical. I just wasn't expecting such a pattern with a 4 game average. I think you are overstating the defenses in these 4 game samples. For example the average of Miami, UT, Miss, Arky last year was 31 points....the average of LSU, UK, Vandy, and UGA was 52.5 points. Was the defenses in defense that large?

It's pretty well-established that it was an offensive change during the Arky game. Demps/Rainey averaged 9 combined carries per game in the first four above (17 vs Arky), which jumped to 14.5 in the next four. Percy's carries dropped from almost 7 to 3.5...

I still think the "bell-curve" of scoring on a season works. In the last two full seasons, the 4th, 8th, and last game averages show this.

[TrumanCoyote]

Percy Harvin's best statistical game last year was the Ole Miss game. It's pretty well established that when the going gets tough the staff has a tendency to go with the stars.

so, yes there might be a correlation of sorts between the overuse of Tebow and a loss, but I don't think you can read a causal relationship into it.

[malscott]

Holy crapola. I knew it was just a matter of time before my math teacher found me online. Cripes. Holy graphs batman! I do appreciate the effort.

My question: What is the probability with some minor adjustments we can continue to win? How do you accurately calculate the fact that other teams are looking to each others successes to adjust against the spread? Is there a parity coefficient in your assumption? Believe it or not, I think Monte Kiffen taught the conference some tricks on D. I personally think a few line adjustments and sticking to the game plan, along with a couple of new plays will change everything and then you'll be too gleeful to work the math anymore. Go Gators...

[rotag]

Great thread. Thanks to all contributors.

[charlegator2]

Great effort, This is exactly the kind of thing that is needed to inform a good discussion and combat the typical analysis that is based on poster's fallible memories and emotion.

I think the focus here, as the OP said, should go to the rolling avg graph of TD% vs TT% and the correlation b/w the same two stats. These actually suggest surprisingly strong relationships to me. Intriguing if true, but I'm a bit skeptical of the methodology because of the size of the correlation.

Are you counting Tebow sacks as a TT? If so, a pass play call on 1st or 2nd down gets represented as a TT when it wasn't and naturally sacks are going to hurt the TD%. A similar case is a designed pass where Tebow scrambles. Again, it becomes less clear if the lack of TD% is due to Tebow-reliant play calling or execution (Tebow decision making, WRs getting open, etc.) Not sure how prevalent these are, so hard to say if they will affect you result.

The sacks issue can probably be cleaned up by publicly available statistics, but probably not the scrambles, you would need video, and I'm not sure you have that much time on your hands . Meyer and the coaches have supposedly been doing the video analysis this week, I'm curious if they have gone back and compared it to prior years like you did. Might be helpful.

I'm also curious if the number of RZ plays in total has been dropping. This could be due to a failure to move the ball more generally (though most are saying that's not happening) and simply an increase in turnovers (which is clearly happening). An interesting analysis would be to look separately at tendencies on each down (1st, 2nd, 3rd) as well as changes in down-and-distance tendencies. Again you probably don't have that time

[PTCGator]

This is a terrific analysis! Many thanks to Matthanuf06 for all your work. Some of the findings are subject to interpretation, but overall, this data documents what we have all been saying. RZ scoring is way off and TT has been less effective lately. Its a combination of a lot of things but one thing is very clear. If the team plays well, we rely less on Tim and we score more in the RZ. I also think we have to account for the fact that we are not scoring many long TD's much this year. That kind of scoring puts points on the board without a play in the RZ. If we can break a few longer plays, then scoring in the RZ becomes less critical. In any case, the bottom line is win the game. Let's hope the coaches have figured out what ails us and have some solutions in store for the rest of the season.

[Matthanuf06]

I think the focus here, as the OP said, should go to the rolling avg graph of TD% vs TT% and the correlation b/w the same two stats. These actually suggest surprisingly strong relationships to me. Intriguing if true, but I'm a bit skeptical of the methodology because of the size of the correlation.

This is the key point. I would not have thought those two things would be so strongly correlated to each other. There are obviously other factors at play so it would be foolish to say it is cut and dried. However in those particular ranges it is pretty clear that the more we use Tebow the less we score in the red zone. I did not think it would be that clear.

Are you counting Tebow sacks as a TT? If so, a pass play call on 1st or 2nd down gets represented as a TT when it wasn't and naturally sacks are going to hurt the TD%. A similar case is a designed pass where Tebow scrambles. Again, it becomes less clear if the lack of TD% is due to Tebow-reliant play calling or execution (Tebow decision making, WRs getting open, etc.) Not sure how prevalent these are, so hard to say if they will affect you result.

There is most definitely human error. I just went off ESPN. If they say it is a sack, then I counted it as a pass. However I'm sure there are some pass plays that Tebow actually ran. However I'm okay with that being in there. I don't want this to be a pure indictment on the play calling, but rather the plays that are actually taking place.

That being said I don't think there is any problems with the conclusions. I don't think the number of errors or mislabels will truly influence it one way or another. The errors should balance out.

My question: What is the probability with some minor adjustments we can continue to win? How do you accurately calculate the fact that other teams are looking to each others successes to adjust against the spread?

We are going 14-0 so no need to worry.

Well I think it is important that we don't look too far into things. I think there are only really two things that we should take out of this.
1. Play calling in the RZ is different
2. The use of TT and our TD% was pretty strongly negatively correlated between 07-08

I do expect that same relationship to continue in 2009, but there are different variables. Like you said perhaps the teams now have a better defensive gameplan and of course we are less talented on offense. Those things have no doubt pushed our TD% higher. However I don't see how those things can break a pretty strong two year correlation.

I think it is safe to say once Lady Luck is on our side and if we begin to call a more diverse game in the red zone that we should convert a couple more touchdowns a game.