Dawg Stats

Tyler Bray or Aaron Murray?

2012-07-12T13:55:00.000-04:00

A bunch of my buddies and some guy named Butch all contribute to a UGA football and UGA sports email listerv called the JunkYardDogs. It's a Yahoo! group, and we recently had a University of Tennessee grad (they practically give degrees out at that hick school) open up a debate about how Tyler Bray was going to dominate UGA this year. It started your standard run of the mill argument about which QB (Bray or UGA QB Aaron Murray) demands the most respect.

I pulled out all of the garbage games (Buffalo, College of Charleston, Norcorss High School, etc) and took the averages for each QB stat. Below is how Bray and Murray stack up...and it's pretty even (click the picture to make the graphs bigger).

Click the picture to make it bigger

As you can see, Bray and Murray are comparable in almost all categories. In my statistical breakdown, Aaron Murray has played in 11 more games than Bray, so be sure to keep this in mind when comparing the two. Both players have similar completion percentages, completions, etc. Bray's attempts went up in 2011, which is partly do to UT having a weak running game in 2011. The jump in pass attempts also correlates to more passing yards per game. Murray has a slightly higher passer rating, but you can see how that rating dropped when AJ Green went professional.

Obviously, the statistics can only tell us so much. By being able to elude the rush, Murray is a little less injury prone and is able to break out of the pocket to try and exploit a broken coverage situation. As another JYD'er mentioned, Bray might have more NFL value due to better mechanics, but short-term value is even between these two QBs.

Tattoo strength aside, the choice between Bray and Murray is a toss-up...but you just can't coach that spike in tattoo strength.

Coach Mark Richt and Coach Vince Dooley

2012-02-07T22:02:00.002-05:00

A lot of people talk about the UGA football team winning ten games on a soft schedule. Sure, Mark Richt has had his fair share of down seasons over the past three or four years, but compared to Coach Vince Dooley there is good reason to be excited.

As you can see, it took Coach Dooley five more years to eventually win a National Championship. So for all the talk of Coach Richt being on the hot seat, I say give him five more years.

Returning Starters

2011-08-07T14:40:00.008-04:00

UGA is officially ranked #22 to start the preseason (Coaches Preseason Poll), so I figured it would be a good idea to try to post something to get us ready for the Chick-Fil-A college kickoff against Boise State. Last fall, I read an article on Smart Football about Boise State and what makes their offense so potent. If you haven’t read it, be sure to put aside thirty minutes to check it out. It’s pretty good reading. Long story short, Boise coach Chris Petersen is a wizard at “maximizing personnel” and creating schematic advantages. I’m not telling you something you don’t already know, but just to put UGA fans on notice, Boise State lights up the scoreboard. Over the past ten years, this is a team that has consistently scored over forty points per game…and there is an upward trend on points scored.

I remember back in college setting up a college football dynasty on EA Sports NCAA Football with UGA around 2001 (Robert Royal was ridiculous in that game, FYI), and after about ten years, Boise State was consistently in the top five. I don’t know if it was a gut reaction or if EA Sports was ahead of the curve, but it was extremely frustrating at the time. Fast forward ten years, and Boise State is indeed a top ten team, so EA Sports was clearly on to something. When handed a Kellen Moore, Chris Petersen must salivate over the playbook each week. He's got a video game offense.

Looking at Phil Steele’s returning starters list from last year, Boise State presents some interesting info; they returned 23 starters on offense, defense, and special teams last year. That’s 23 starters who knew the entire playbook for two years in a row. It also created a serious advantage for someone like Chris Petersen – offensive wizard – because he can open up the playbook a little more. Last year, UGA returned 18 starters to win six games. One of those guys was AJ Green, so really the Georgia football team was without a full-on offensive creature for the first four games. Obviously, his non-presence will be felt this year, but UGA has to find a way to win without him AND a veteran running back.

(SKIP THIS NEXT SECTION IF YOU DON’T GIVE A CRAP ABOUT MATH)

I ran a regression to determine whether the number of returning starters will affect the number of wins that a team has had over the past two years. I removed MAC and Sun Belt Conferences from the data because those teams are perennial doormats for BCS conference teams. I left in Conference USA (UCF killed us in the bowl game, so they must be pretty legit). I also left in the WAC (Boise State) and Mountain West (TCU) – welcome to the big leagues, guys. Basically, I took the wins for every team in the NCAA from 2010 and 2009 and used the number of returning starters to try and determine whether the amount of returning starters will have an effect on the win totals. Most statisticians prefer to use a 95% confidence rating (alpha). I used 90% because it's tough to find much significance in football stats. Using 90% confidence rating, regression shows that the number of returning starters does affect the number of wins that a team has in a given season.



F	Significance F
2.83989629	0.093549763



P-value	Lower 95%	Upper 95%	Lower 90.0%	Upper 90.0%
2.27055E-05	2.771065183	7.384085081	3.144721721	7.010428543
0.093549763	-0.022643683	0.288556395	0.002563654	0.263349059

Since the returning starter data is significant, we can determine the expected amount of wins from both UGA and Boise State – depending on the number of starters they return this year:

Regression Statistics
Multiple R	0.11981034
R Square	0.014354518
Adjusted R Square	0.009299925
Standard Error	3.000332951
Observations	197

ANOVA
	df
Regression	1
Residual	195
Total	196

	Coefficients
Intercept	5.077575132
RET STRTERS	0.132956356

(OK, YOU CAN START READING AGAIN NOW)

Here is the expected number of wins – depending on the number of starters returned in 2011, using information from every team in the NCAA’s returning starters over the past two seasons. This is the number of wins that a team can expect to have based on the number of starters returning on offense, defense, and special teams combined:

# Returning PREDICTED WINS

10	6.407138695
11	6.540095052
12	6.673051408
13	6.806007764
14	6.93896412
15	7.071920477
16	7.204876833
17	7.337833189
18	7.470789546
19	7.603745902
20	7.736702258
21	7.869658615
22	8.002614971
23	8.135571327
24	8.268527684

UGA returns 17 starters and Boise State returns 14, so we should win 0.4 more games than Boise State this year. This is with 90% confidence, and only relies on past data. Obviously, I don’t expect Boise to ONLY win almost 7 games this year because they play in the W(h)AC(k) conference. Boise will win ten games easy with that QB and those schemes…but the SEC isn’t quite as clear.

There is a legitimate chance that UGA only wins 7 games this year. We have unproven running backs, need someone to step up at WR, are thin on OL, and have many new faces on defense. Vanderbilt and Kentucky aside, it will be interesting to see how close these respective teams are with their end rankings at the end of the season.

CONF	2011 Team	# Returning Starters (Off/Def/ST)	Projected Wins
SEC	Vanderbilt	21	7.869659
SEC	Georgia	17	7.337833
SEC	Kentucky	17	7.337833
SEC	Alabama	17	7.337833
SEC	Mississippi	16	7.204877
SEC	Arkansas	15	7.07192
SEC	Miss St	15	7.07192
SEC	LSU	15	7.07192
SEC	South Carolina	13	6.806008
SEC	Florida	13	6.806008
SEC	Tennessee	13	6.806008
SEC	Auburn	7	6.00827

Now, I’m not going to sit here and type that Vandy and Kentucky aren’t worth a damn, but this data doesn’t really include much information about strength of recruiting, coaches, or other factors that affect team wins (AJ GREEN, TERRENCE CODY, CAM NEWTON, et al). So it’s probably not likely that Florida will only win 6 games. Bear in mind that this data includes the entire NCAA FBS, which includes boring teams like anyone in the Big Ten or Big East. Most SEC teams pummel a few D-1A schools and have inflated win totals at the end of the year, but it will certainly be interesting to see if a team like South Carolina isn’t that great.

So what does this tell us? Teams are better when they return more starters. What else? Kickers, punters, and return men are an important part of a team’s win totals. When I ran the regression without special team starters coming back, the data wasn’t significant. So we can’t say with certainty that win total only comes from positional players returning. Special Teams helps teams win games, “significantly”. Hopefully, this works to our advantage comes labor day weekend in the Georgia Dome.

Feel free to suggest new articles in the comments section.

What is Regression? Do stats matter?

2011-06-30T22:30:00.006-04:00

Many people reading this blog may not have a clue about regression or statistical analysis. There are literally thousands of websites online that can explain linear regression a million times better than I can. So before getting into the heart of the blog, let me explain a few things about regression:

What is regression?

One cool thing about regression is that we can predict whether the numbers state that UGA will win or lose a given game based on the information we plug into the statistical model. Regression is great at making confident predictions about the future. For instance, if we want to see whether UGA wins or loses a game when winning the opening coin toss then regression can make us more confident in whether they will or will not lose.

Does regression work?

Well, that's what this blog looks to find out. Do I have any idea what I'm trying to find? Not really. But after a period of trial and error, I'd like to think we can come up with sound reason to support or refute why Mike Bobo should be fired or why we should run the ball more out of 4-wide sets.

Do the stats matter?

Sure. If stats didn't matter then statisticians wouldn't keep track of them during a game. Simply put, we all know why the uga football team succeeds or fails during a given week. But maybe there are underlying pieces of data that show we played better or worse than what we saw as fans in the stands. Yes, stats matter, but no they are not the end-all-be-all of football. Football is a zero-sum game: someone will win or lose. Often times there will be a huge play that makes or breaks a game, but this blog's purpose is to identify key stats to give the casual Georgia football fan something to watch on Saturdays.

What qualifies me to write about this?

I have no qualifications for writing this blog other than the fact that I know how to run regression, check for correlation, forecast data that trends upward or downward, forecast data with or without seasonality, and keep accurate records of the offensive play calling. Other than that, don't expect me to break out the sabermetrics. Don't get me wrong, I love the geeky stuff, but ultimately, this blog has to be entertaining for someone who literally knows nothing about stats.

That said, my next post should have some good data to review.

First Post - Welcome

2011-06-15T21:36:00.000-04:00

Hello,

Welcome to Dawg Stats. My name is Geoff and I am a graduate student by night and banker by day. Being the UGA Football dork that I am, this space will be used to run statistical regression and other mathematics for determining why UGA has been successful or (as of recently) mediocre at football. Throughout the season, I will try to break down some games using statistical analysis - NOT the stuff in the box score. Basically, the hard hitting stuff.

Feel free to follow or add the RSS feed to be updated when I get around to posting some data. My goal is to take a deep dive into the weekly or seasonal statistics and identify enablers for success or failure.

This could all be a giant waste of time, but I've got a baby on the way, attend grad school, and work 40+ hours per week - so why not add some more to the plate?