Sep 20, 2014

ND Fan Radio Appendix 9-17-14: Turnover Margin

It’s just like riding a bike…virtually….with words and numbers…and no physical exertion. As I settle back into the old digs at John Lynch’s Dad, I plan on continuing much of the same analysis I’d previously been doing for Down the Tunnel. Less there be any confusion, I think y’all should still be checking out Down the Tunnel's content, but with that out of the way, new opportunities have also presented themselves.

One of those opportunities was having an opportunity to go on to ND Fan Radio’s call in show on Wednesday night. You can find all of their stuff on the iTunes store through The New ND Nation's feed. Additionally, you should follow both of the hosts on Twitter if that’s your thing. The guys are @IrishTightness and “Air McFly” @NDFanRadio to keep up to date with their comings and goings.

The hope will be to get on more often than not and do some deep dives (or shallow dives depending on how we’re feeling…perhaps just debating the value of 1990’s Pop Rock?) on some statistics that I found interesting or are worth noting. That of course means I’ll be speaking in a lot of numbers in a short period of time. Speaking in numbers is good for computers (Siri excluded), but not so good necessarily for digesting what was said in 10-15 minutes. With that in mind, I welcome you to the first edition of the “ND Fan Radio Appendix.” Following any appearance I make during their 9:30 PM ET Wednesday call-in show, I’ll try to post a quick run through of the stats I talked about. The goal will be to shoot for Fridays, but you can see how well that’s started out.

Enough of this promoting, explaining, and wording. Let’s get to numbering.

Stat of the Week:  Through Notre Dame’s first 3 games, they're averaging a +2.7 Turnover margin per game. That’s good for a top 5 ranking in college football year to date.

Explanation:  Turnover margin per game is a straightforward concept. Nothing deep about it but always worth explaining before we get into it. Turnover margin is defined as “Turnovers created by Team A – Turnovers surrendered by Team A.” A positive margin means a team creates more turnovers than it gives up. A negative margin means the opposite.

Obviously, positive means “good,” and negative means “bad.” It’s effectively the single number method of telling you “who won the turnover battle”…a phrase loved by analysts and fans alike. And, in a rare instance of clarity, for good reason. Turnovers and more importantly turnover margin impacts so many other stats that come to decide the result of a game. A turnover deep in your opponent’s territory can derail an otherwise promising offensive drive (see: Crist, Dayne; South Florida). It harms other secondary stats that let you know how an offense is performing such as yards per point, time of possession, and scoring efficiency. Yes, turnovers may be just one facet of a game and somewhat fluky, but they’re also remarkably important and quite predictive of how a team will fare, usually. More on that later.

Stepping away from the stat world for a second, anyone who’s watched a dramatic game, drank in agony when a goal line interception shatters dreams, or quite simply watched the Tommy Rees era of Notre Dame football understands the emotional impact turnovers have on the feel of the game too. Whether true or not, there’s a perceived psychological advantage to turnovers that can’t be dismissed altogether.

What I’m Curious About: So, the real question is: How good is a +2.7 margin? Sure, it’s top 5 right now, but will that continue? The answer is: Yes, A +2.7 TURNOVER MARGIN PER GAME IS GOOD….REAL GOOD…..LIKE, YOU’RE NOT GONNA CONTINUE THAT GOOD, GOOD.  To show you what I mean, the chart below displays the turnover margin per game for the team that finished first in the nation at season’s end dating back to 2005:

Year
2013
2012
2011
2010
2009
2008
2007
2006
2005
TO Margin
1.7
1.8
1.8
1.8
1.7
1.8
1.8
1.3
1.8

Whether you’re good at logic puzzles or not, I think you can seen the pattern. Basically, a +1.8 is about the top of the mountain when it comes to turnover margin per game over the course of an entire season.  Basically, just by regression we should expect Notre Dame’s turnover margin per game to drop off by at least 1 turnover created per game. That’s okay and is not necessarily a concern. While that number may come down, maintaining a turnover margin on the positive side of zero over the course of a year is still a huge advantage. In the Brian Kelly era, the turnover margin per game numbers have been a beacon of what the season will come to be.  The next chart shows each of Brian Kelly’s first 5 seasons with their turnover margins through the first three games, what their record was, and where the team finished:

Year
TO Margin First 3
Record First 3
Point Differential First 3
End of Year Record
2010
-1.3
1-2
4
8-5
2011
-2.7
1-2
11
8-5
2012
1.67
3-0
60
12-1
2013
0.33
2-1
18
9-4
2014
2.7
3-0
78
???

I apologize in advance for doing this to everyone, but if you’re asking me based off of these measures, what does the 2014 start look most like, it looks most like a ramped up version the 2012 start. When on with ND Fan Radio, both Air McFly and IrishTightness pointed to other areas where this start feels like 2012 including total rushing yards and scoring defense. It’s not worth drawing any hard parallels to the national title run of 2012 just three games in, but hopefully this helps confirm for you that you were in fact not crazy with those special tingly feelings. Also, put down the glass of whiskey.

The other thing this chart points out is how there is a loose but noticeable correlation between turnover margin per game and point differential which is what I alluded to earlier. Turnovers create (or take away) scoring opportunities and a positive margin more often than not leads to a subtle but difficult to overcome advantage of more possession and opportunities to score.

The Brian Kelly Era:

That takes me to the pet project from the week. I set out to figure out what “winning the turnover battle” has meant in the Brian Kelly era. The issue with margins/per game in the early going is that outliers can impact the number and you don’t necessarily know whether a team is consistently winning or losing the turnover battle on a game by game basis. A team with a +1, 0, 0 would have the same turnover margin per game as a team that had +5, -2, -2 through three game. Of course, the first team would be much more likely to have a good record than the team that had deficits of 2 in 66% of their games.

Therefore, I broke down Brian Kelly’s performance by turnover margin situation for each game during his 4+ years. The results are as follows:

Situation
Games
Wins
Losses
Win %
ALL
55
40
15
72.72%
Margin > 0
25
24
1
96%
Margin = 0
7
6
1
85.71%
Margin < 0
23
10
13
43.47%
Margin ≥ 0
32
30
2
93.75%

So what I’ve got highlighted is what I tweeted out on Thursday: Since Brian Kelly has been at Notre Dame, he is 24-1 with a positive turnover margin and 30-2 if turnovers are even or better. The evidence was overwhelming: an amazing predictor of success for Kelly’s teams was simply to look at turnover margin. If it wasn’t in the red, there was a 96% chance we’d won that game. Great, fantastic, awesome, but CONTEXT!! I really had no idea how to absorb this information. It sounds impressive, but where does it stand in relation to other college football coaches?

With a substantial amount of assistance from twitter colleague and masochistic stat nerd @andrewwinn (please follow him if you’re not already doing so), we can provide a little more substance to this topic. The question was two fold: 1) What to look at, and 2) Who to look at. One proved easy to answer. I wanted the data for other college football coaches since the beginning of 2010 when Brian Kelly came to Notre Dame. I also wanted to exclude any match-ups which might’ve occurred with FCS opponents. The year restriction had one exception: I was asked how Kelly matched up compared to other ND coaches. With that in mind, Charlie Weis’ numbers were also crunched through 4 seasons and 3 games during his ND tenure.

The who proved more difficult. I was curious about coaches with similar records, better records, and slightly worse records. I preferred coaches who were at one school for the entire time period, though one partial exception was made. Truth is, if a coach is much worse at winning than Kelly, he probably doesn’t make it 5 years at one school. With that in mind, the sample ended up including two of the winningest coaches of the past 4+ years (Nick Saban and Chris Peterson), several coaches who have substantially similar records to Kelly (Mike D’Antonio, Bo Pelini, and Mike Gundy), two with slightly worse records (Mark Richt and Art Briles), and then Charlie.

For the comparisons, I present two charts. The first one will show each coach’s winning percentage by turnover margin category:

Coach
Total
> 0
= 0
< 0
≥ 0
Saban
86.53%
96.55%
75%
73.33%
91.89%
Peterson
83.01%
88.46%
78.57%
76.92%
85%
Gundy
76.92%
96.55%
61.54%
40%
85.71%
D’Antonio
75%
88%
70%
58.82%
82.85%
Kelly
72.72%
96%
85.71%
43.47%
93.75%
Pelini
69%
94.11%
90%
44%
92.59%
Briles
68.5%
83.33%
81.81%
33.33%
82.85%
Richt
63.46%
88.46%
63.63%
20%
81.08%
Weis
58.49%
75%
40%
40%
65.79%

The coaches are arranged by overall winning percentage. As you can see, Brian Kelly’s right in the middle of the pack. The numbers revealed that Brian Kelly is in fact elite at converting an even or better turnover margin into a win. None of the other eight coaches reviewed since 2010 had a higher win percentage in such situation. Only Nick Saban and Bo Pelini even broke the 90% threshold with Kelly.

Not surprisingly, the two coaches with the truly elite overall winning percentages did something the others could not: overcome a turnover deficit to get a win. Nick Saban and Chris Peterson were the only two coaches to break the 70% win percentage when having a turnover deficit. In fact, they were the only two coaches to break 60%! On the opposite end of the spectrum, there was Mark Richt whose winning percentage of 20% with a turnover deficit was by far the worst. Five of the nine coaches had winning percentages with a deficit between 40 and 60 percent. Richt’s frustrations with breaking through in the SEC East probably have a lot to do with so rarely being able to overcome a sub-par turnover game.

And then….there’s Charlie. Charlie Weis was the only coach not to break an 80% win percentage with a positive turnover differential. He was also the only coach to lose more games than he won with an even turnover margin. As if that’s not devastating enough, the chart below will really hammer home why that was so painful to the Charlie Weis era.

The next chart shows the percentage of games each coach found his team as a percentage of total games played from 2010 – present (excluding FCS competition):

Coach
Total Games
> 0
= 0
< 0
≥ 0
Saban
52
55.77%
15.38%
28.84%
71.15%
Peterson
53
49.06%
26.42%
24.53%
75.47%
Gundy
52
55.77%
25%
19.23%
80.77%
D’Antonio
52
48.08%
19.23%
32.69%
67.31%
Kelly
55
45.45%
12.72%
41.82%
58.18%
Pelini
52
32.69%
19.23%
48.08%
51.92%
Briles
50
48%
22%
30%
70%
Richt
52
50%
21%
28.84%
71.15%
Weis
53
52.83%
18.87%
28.30%
71.7%

The two coaches who won the highest percentage of their games when getting and even or better turnover margin were also by far the worst at their teams making it to that situation. If there’s a reason Brian Kelly’s teams haven’t been even more successful, it’s that his particular mix of offensive turnovers and lack of defensive turnovers placed the team in a turnover deficit more frequently than any coach other than Bo Pelini. Probably also goes a long way towards explaining why so many Nebraska fans are growing tired of Coach Bo.

Mike Gundy was the only coach to have his team at even or better over 80% of the time and explains why he has the same number of wins as Kelly over the same time frame with fewer losses.

That brings us to Charlie. In terms of wasted opportunities, Charlie Weis won or at least tied the turnover battle more frequently than any coach reviewed other than Mike Gundy and Chris Peterson. He even edged out Nick Saban in that respect. However, his teams performed so much more poorly in those situations that any possible advantage was negated. Had he performed just at the norms, there’s a non-zero chance Charlie Weis might still be honoring that 10 year contract Notre Dame gave him. In a way, we all owe Charlie a degree of thanks for being less successful at leveraging turnover advantages into wins.

Kelly’s Quarterbacks:

Finally, I think many Notre Dame fans would point towards the failed (or at least tepid) tenures of Dayne Crist and Tommy Rees as an explanation for why the team has not more frequently found itself in an advantageous situation that could be exploited. The numbers bear this out quite favorably. Suffice it to say that Everett Golson has been the best quarterback in the Notre Dame/Kelly era. His legs and improvisation are a definite plus. He’s also been less turnover prone. Between 2012 and 2014, Everett Golson has played in 15 games (he did not play in 2012 versus BYU). In those games, there is a distinct turnover advantage which developed compared to other quarterbacks:

Quarterback:
Games
% ≥ 0 Turnover Margin
% < 0 Turnover Margin
Golson
15
80%
20%
All Others
40
50%
50%

If the goal is converting Brian Kelly’s superb track record of converting even or better turnover margins into wins, Everett Golson’s play has certainly assisted in that respect. It’s still a reasonably small sample size, but in the 15 games in which Golson has played, Notre Dame has an average Turnover Margin per Game of 1.0. In the remaining 40 during Kelly’s tenure, the average Turnover Margin is -0.3. Let’s all hope these trends can continue.

Please let me know what questions you might have or what else you’re interested in learning about. Hit me up on Twitter at @IrishMoonJ, and I hope you’ve enjoyed.

- Moons