Mark Richt's Coaching Record at Georgia
by Year2 on Apr 28, 2009 11:00 AM EDT
in SEC Football
Email
PrintLast offseason, I took a look at various coaches' records on my personal site. This year, I'm going to go through the SEC coaches with a couple years of conference head coaching experience, updating anything I did last year and adding some more to it.
It's going to go in a pseudo-random order, and first up to bat is Georgia's Mark Richt.
With 82 wins to his name, Georgia's Mark Richt has the potential to be the winningest coach in Bulldog football history. He's just 49 years old, and given his average clip of about ten wins per season, he will pass Vince Dooley at around age 61 or 62.
He's drawn criticism every now and then for not winning a national championship, but his 13-1 team in 2002 would have won a national title in nearly every year of the BCS except that one. More pressing is his 2-6 record against rival Florida, but he's also a combined 12-4 against his other rivals Auburn and Georgia Tech.
The following stats about his record do not include his five wins against I-AA teams, so keep that in mind when the win totals don't add up to the full 82.
This is what Richt's record looks like broken down by site:
Surprisingly, Richt actually does better on the road than he does at home. The neutral site record consists of Florida-Georgia games and his SEC title game appearances. We can also see a good track record at getting his teams ready for bowls.
This is what Richt's record looks against four tiers of opponents. The first tier is teams with end-of-year winning percentages of at least .750. The second tier is from .500 to .749 and so on.
Funnily enough, his teams average more points against second tier teams than third tier. We can also see that, though his record is a few games below .500, his teams actually outscore his first tier opponents on average.
If you're curious, the one loss to a third tier team was the 2006 loss to Vanderbilt.
Now comes the part that is new for this year: a year-by-year look at each coach's so-called Pythagorean wins. I learned of the Pythagorean expectation from Roll Bama Roll's outsidethesidelines, and he applies it to conference games each season. For this, I included all I-A games each year.
Basically, Pythagorean expectation tells about how many games a team should have won given the number of points for and allowed, and it is usually very accurate. I used the exponent 2.37 that Pro Football Reference divined instead of just two. If you're curious about the exact formula, hit that link in the previous paragraph.
The difference between Pythagorean wins and real wins is often termed "luck," but "chance" is probably a better term, and it can also be considered a measure of if a team overachieved or underachieved in a season. Here's Richt's Pythagorean wins by year:
Richt's 2002 team overachieved by more than a full game relative to expectation by either being good, lucky, or a combination of the two. There's absolutely nothing wrong with that either because often times it takes a season where the stars align for you to end up with a 13-1 (or similar) record. Just look at the 2003 team which actually had more Pythagorean wins but that ended up 11-3 thanks in large part to having to play 13-1 national champion LSU twice (both losses).
Differences of less than a game aren't all that much to get worked up about, but look at the last two years. Georgia overachieved by more than a game in 2007 and over a game and a half in 2008. It goes to show what offensive line injuries and defensive breakdowns should have done to the Bulldogs, and what a great job Richt, Stafford, and Moreno did to overcome them.
Since Pythagorean wins are often a good indicator of the future, it looks like 2009 could be a rough year for the rebuilding UGA program. They haven't broken the nine-win mark since 2005. However, Richt has overachieved more often than he hasn't, so it's probably a good bet that Georgia will win a game or two more than it should.
Fancy stats aside, it's hard to make the case that Richt has been anything other than a runaway success in Athens. He's a genuinely good guy who generally runs a clean program. He also happens to have far and away the best winning percentage of any Georgia coach that stayed for longer than three seasons.
The Georgia job is his for as long as he wants it, and it's hard to see him going anywhere else. Mark Richt and Georgia are a near-perfect fit, and the two will be together for quite some time into the future.
7 comments
|
0 recs |
Do you like this story?
Comments
Nice post
On the chart showing Pythagorean wins by year, are you counting bowl games? If you aren’t (and I take it that you’re not), the real wins for 2003 are wrong. If you are, the real wins for 2002, 2004, 2006, 2007, and 2008 are wrong. Since it’s more likely that you miscounted one year than five, I’m assuming you’re not counting bowl games, which causes me to wonder why not. Is it because you’re only including games scheduled in advance? If so, why are S.E.C. championship game appearances apparently being counted?
You indicated that Pythagorean wins have predictive value. How is this determined? Since the formula takes into account points for and against, how can we determine in advance how many games a team “should” win? Is it based on last year’s numbers? If so, how reliable is that? Georgia lost two first-round draft picks from its 2008 offense. What predictive value do last year’s offensive numbers have for this year’s Bulldogs?
Finally, how reliable is the points for and against calculation as an average? For instance (and I use this as an example, not as an indictment), Urban Meyer tacked on a field goal in the final minute or two of a game against Miami that was already decided last year, whereas Mark Richt routinely has his quarterbacks take a knee in that situation.
Without making any value judgments as to which is the better course, it represents a purely statistical difference that is reflected in the Pythagorean formula. If Mark Richt orders his team to take a knee in a situation in which the Bulldogs could kick a field goal, but Urban Meyer kicks the field goal in the same situation, aren’t the Gators getting credit for numbers which are utterly inconsequential in determining wins and losses? In short, are style points being allowed to influence mathematical calculations upon which they have no actual bearing?
These are genuine questions, not challenges, mind you. This was an interesting and fair piece and I’m intrigued by (albeit suspicious of) the application of Bill James’s baseball concept to the very different sport of football.
Go 'Dawgs!
The totals leave out I-AA games (not bowls) so the missing games are ones against Northwestern State (2002), Georgia Southern (2004 and 2008), Western Kentucky (2006), and Western Carolina (2007).
As I understand it, Pythagorean expectation came about observationally, though someone has gone about mathematically proving that it should be close to reality.
It’s based on a probabilistic view of games – if a team scores exactly as many points as they give up, meaning they score half of the total points in their games, then you would probably expect that team to win about as many as they lose. From there, the greater percentage of the total points in their games they score, the more games you’d expect they would win. People with far more math in their background have gone and figured out the exponents in the basic Pythagorean wins formula that work best are about 1.81 for baseball, 2.37 for football, and either 14 or 16.5 for basketball. So, people have tried to customize it for each sport.
As for it’s predictive power, that’s also more of an observational thing than an mathematical proof. For instance, in 2007 OTS picked out Tennessee (6 SEC wins versus 3.94 SEC PyWins) and Miss State (4 SEC wins versus 2.58 SEC PyWins) as overachievers who would fall and Ole Miss (0 SEC wins versus 1.40 SEC PyWins) as an underachiever who would rise up.
A lot of people pegged MSU and Ole Miss as candidates to fall and rise in ‘08 without those, but almost no one predicted Tennessee’s dramatic fall. Going back a year, the ‘06 Gators won 7 SEC games on the back of 5.55 SEC PyWins, and the Gators fell back to 5-3 in conference in 2007. Furthermore, LSU’s PyWins had declined each year since 2005 (despite the ‘07 BCS title) and Auburn’s had declined each year since 2004 leading into last year’s disappointing performances for each team.
In Georgia’s case 2002, 2007, and 2008 stick out as years where chance favored the Bulldogs (differences of under a game aren’t significant). Since chance generally evens out over time, one might expect a step back in wins the next year. UGA did step back a win in both 2003 and 2008, so in that sense the PyWins did have predictive power. It would also say that UGA won’t do better than last year’s 10-3 mark. It doesn’t tell you how far a team will probably rise or fall, it just gives a likely direction.
And you are right that running up the score can have an effect, though adding a spare field goal or touchdown here or there doesn’t make that big a difference. For instance, Arizona’s PyWins for 2008 say they should have been about a 12 win team, until you take out the 70-0 win over Idaho at which point they drop back to 7.08 PyWins (and they won 7 if you take out the Idaho drubbing). I know some people have looked into when and where to cap blowouts in college football, but that work is still ongoing as far as I know.
In my own research into it, I have found taking out I-AA games gives a pretty accurate win percentage except in the case of teams with one obvious outlier that doesn’t get evened out (like Arizona). There will never be a formula that is perfectly accurate since chance will always be a part of the game, but this one comes pretty close all things considered.
Correction
That should say that raw PyWins (with Idaho included) have Arizona as a 10 win team in 2008, not 12 win.
Thanks for the clarification
Sorry I missed the point about Division I-AA teams, which you were right to remove.
I’ll be interested to see how this plays out with other coaches, particularly Urban Meyer.
I have a theory, which I set forth here, that, generally speaking, college head coaches fall into one of two categories. There are the big game coaches like Pete Carroll and Urban Meyer, who, whether through intense enthusiasm (Coach Carroll) or intense focus (Coach Meyer), can always prepare their teams for big games. A year ago, the news media made it clear that Southern California’s biggest game was against Ohio State and Florida’s biggest game was against Georgia. Both coaches circled those dates on the calendar, and both were utter annihilations.
However, no one can be that intense, that focused, and that enthusiastic all the time, which is why coaches like Pete Carroll and Urban Meyer, when they slip up, tend to slip up against lesser competition; e.g., U.S.C.‘s losses to Stanford and Oregon State, Florida’s losses to Auburn and Ole Miss under Urban Meyer, and Florida’s losses to Syracuse and Mississippi State under Steve Spurrier.
On the other hand, there are the coaches like Vince Dooley, Mark Richt, and Mack Brown, who maintain an even emotional keel and never get either too high or too low. These coaches split the toss-ups and almost never lose games they ought to win.
In the case of Coach Richt, I believe the numbers you cite bear that out. Mark Richt is 9-12 against top-tier teams and the games are close (24-23), which supports my idea that he is the sort of coach who splits the toss-ups. Even excluding Division I-AA opponents, his teams are 30-1 against opponents below .500. I’d be willing to bet that Coach Meyer has a significantly better winning percentage against first-tier teams and a significantly worse winning percentage against second- and third-tier teams.
We shall see, though. In any case, thanks for running these numbers. It’s comforting to see that Mark Richt’s teams have never significantly underperformed yet have overperformed agreeably often. It’s also comforting to know that, in years after overachievement relative to Pythagorean wins (2008 did not feel like an overachieving year), his teams have dropped off only marginally, notching one less win against Division I-A competition the following year.
That would put the Bulldogs at eight real wins against Division I-A opposition in 2009, plus a victory over Division I-AA Tennessee Tech. That would give Georgia eight straight seasons of nine wins or better, which would be an active streak no other S.E.C. team could claim. In the long run, as a fan of a team whose winningest coach won his first national championship in his 17th season, I have to believe that slow and steady wins the race. I’d darned sure take nine wins next year if I could lock that up right now, considering the players we lost and the slate we face.
Go 'Dawgs!
Meyer will be coming up on this series, but I did run Carroll’s numbers last year and here they are updated through 2008:
Home: 45-3 (.938)
Away: 37-10 (.787)
Bowl: 6-2 (.750)
Total: 88-15 (.854)
By tiers:
First: 18-5 (.783), Avg Score 36-19
Second: 31-8 (.795), Avg. Score 31-17
Third: 30-2 (.938), Avg. Score 39-16
Fourth: 8-0 (1.000), Avg. Score 48-8
Note: one of the third tier losses was in his first year when USC went 6-6, the other was 2007 Stanford.
PyWins:
2001: 8.44 PyWins, 6 real wins, difference of -2.44
2002: 10.76, 11, 0.24
2003: 11.32, 12, 0.68
2004: 12.06, 13, 0.94
2005: 11.18, 12, 0.82
2006: 10.91, 11, 0.09
2007: 10.97, 11, 0.03
2008: 12.57, 12, -0.57
It looks as though from PyWins that Carroll has always been right on target or a bit above. That first year was a disaster, but it’s been gravy ever since. It goes to show how difficult getting to 12 PyWins in 13 games is – it took his undefeated, undisputed national champion in 2004 and his all-timer of a defense in 2008 to crack that level.
So in this sense, it should not be surprising that USC loses a game or two each year. It’s who the Trojans lose to that is the surprise: it’s the kind of blase 7-9 win teams (and that ’07 Stanford squad) that get them. That plays right into your theory about Carroll, and it also helps that four of his 5 top tier losses came in his first two seasons. That leaves the Vince Young Game as his only top tier loss after his program hit its stride.
I understand the process of looking back to see if a team was “lucky” or “unlucky” relative to their points scored. I even get how looking at a baseball team halfway through a 162 game season might give insight into second half results to look for.
But I fail to understand the predictive value for football in that it isn’t measuring the same team when you talk about the following season. Yes, if no one graduated or got hurt and you played the same offense and defense, it should even out in the long run. But if you graduated a bunch of seniors or have an incoming class of outstanding sophmores and freshman recruits (why yes, I am an Alabama fan, why do you ask?) how do last year’s numbers have a bearing on this year’s performance?
This blog post looks at some historic data in relation to how Pythagorean wins do at predicting future performance.
What the measure of Pythagorean wins does is say how many wins are likely for the given amount of points for and points allowed. If there is a significant discrepancy between the predicted and actual wins, then it’s likely that chance was noticeably in or out of your favor that year. Good or bad luck, if you will.
Chance and luck tend to even out over time, so if a team is extraordinarily lucky or unlucky in a particular year, it’s unlikely they will continue to be in the next. It’s not impossible of course – just look at the ’08 Bulldogs – and transformational changes (like that mass exodus of seniors or a blowout recruiting class) can overwhelm the chance factor.
Another way of looking at how chance can even out from one year to the next is something Phil Steele does in his preview magazine every year. He looks at teams’ performances in close games (final margin of seven points or less) and picks out teams that won or lost a net of 2 or 3+ games. The idea is that teams that win significantly more close games than they lose (or vice versa) probably were the beneficiaries (or victims) of chance. Note the word “probably;” we’re dealing in likelihoods, not absolutes.
In the last six years, teams with 3+ net close losses have been the same or improved the next season 72% of the time. Over the same span, teams with 3+ close wins declined 71% of the time. There are obviously going to be more explanations than just luck evening out as to why those teams improved or fell off, and 3 out of ten bucked the trend.
Still, it illustrates the point. Chance evens out over time, and that can be useful in projecting how a team does in the future.
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
