clock menu more-arrow no yes

Filed under:

Advanced Stats Can’t Measure LSU’s Offense

New, 44 comments

The limits of analytics OR Sometimes, total offense DOES matter.

Kent State v LSU
Beckham watches Jarvis make a great play
Photo by Stacy Revere/Getty Images

The most common complaint among LSU fans over the past few seasons has been the lack of offensive production. It wasn’t just that the team kept losing three to four games each year, it’s that the team lost them in seemingly the same way.

My kingdom for a quarterback, right?

However, advanced stats have been rather kind to the LSU offense. Whenever someone comes to the site to complain about total offense, we kindly remind this person that total offense isn’t all that useful, and refer them to yards/play which, shockingly, LSU has a decent track in. This is a self-serving argument, and one I’m guilty of.

I don’t want to throw out all advanced stats. Bill Connelly, especially, is doing the Lord’s work and greatly increasing our understanding of college football. But advanced stats are still in their infancy with regards to college football, and even if they were 100-percent perfect, they would still miss things simply by the nature of the short season.

Bill wrote a terrific piece this week on the nature of big plays. You should read the whole thing, but to summarize it (poorly), big plays are the single most important factor in winning games, but the best way to create big plays is to be efficient. There’s a bit more to it than that, but explosiveness is generally unpredictable, so the best way to create explosiveness is to keep the offense on the field. And teams do that by staying efficient.

Bill’s got a lot of charts, and y’all know I love charts. He also makes a compelling comparison to Ken Pomeroy’s research on three-point defense in college basketball. Once the shot is in air, the defense has little control on whether it goes in, so the key to a good three-point defense is not letting the other team take three pointers.

Here’s the rub. These assumptions, which I think are pretty good when applied at the macro level, fall apart at the micro level of looking at LSU. LSU keeps churning out good ratings according to the stats, but let’s be frank, the advanced stats are wrong. LSU has not had a productive offense.

So what on Earth is going on that makes LSU immune to stats analysis?

Let’s first identify the purpose of an offense. The goal of an offense is not to be efficient. It is not to have explosiveness. It’s not even to gain yards. The goal of an offense is to score points. You win games by scoring more points than the other guys.

Now, getting yards obviously helps. So does good field position, which is out of an offense’s hands, which is why we want to look at yards, to isolate the offensive contribution. However, when looking at yards, you have to remember that we’re looking at a component part. A really important component, but a component nonetheless. The goal is points. And LSU, this decade, has finished in the top 25 nationally in points just twice (17th in 2011 and 23rd in 2013).

Despite a middling scoring record, S&P+ loves the LSU offense. The advanced metric has ranked LSU as a top-25 offense in three of the last four years, and four of the past six. How is this even possible? Are the stats wrong?

The stats aren’t wrong, really. It is designed to rate efficiency and explosiveness, yet explosiveness is a function of efficiency. The formula is bound to rate efficient teams highly, simply by design. Now, this is where I will part from the advanced metrics, as Bill uses Success Rate and IsoPPP to build his ratings system. However, he’s only published those stats going back three years, so I’m going to use some traditional stats as stand-ins: yards per play for efficiency and 20-plus yard plays for explosiveness.

LSU has been top 25 in YPP in three of the past four years and only once in 20-plus yard plays (2013). Not surprisingly, those are the same three seasons in the past four that S&P also has LSU as a top 25 offense. LSU has been efficient, gaining a bunch of yards per play, but the explosiveness and more importantly, the points, haven’t come.

Is LSU just the unluckiest team on earth? If it were one year, then you could make the case, but this has been going on for half a decade. There’s clearly something else at work here. Something Bill hasn’t built into his formula because it usually isn’t a huge factor, but it is in LSU’s case.

Let’s put all of those ordinal ranks in one chart, add a column or two, and maybe we can see the issue:

LSU offense 2010-2016.csv

Year YPP 20+Yd Play YPG Plays TO Lost PPG S&P
Year YPP 20+Yd Play YPG Plays TO Lost PPG S&P
2010 73 80 86 77 71 45 35
2011 50 67 86 63 1 17 6
2012 75 56 87 70 27 57 37
2013 7 5 35 90 41 23 13
2014 62 48 80 71 27 76 58
2015 17 38 39 120 2 43 14
2016 13 62 59 125 35 68 22

It’s about pace and turnovers.

Let’s address turnovers first because that column looks like a positive. And without context, it is good. LSU doesn’t commit a bunch of turnovers. LSU hasn’t eliminated turnovers, but the offense is routinely one of the best in the nation at protecting the football. That sounds great, Poze, so what’s the problem?

The problem here is philosophical. The goal of an offense, as we mentioned above, is to score points. But LSU is using its efficiency not to drive down the field and light up the scoreboard but to instead prevent turnovers.

Know who turns the ball over a lot? Teams that take risks. (okay, and bad teams, so let’s start with not being bad at offense). Check out some of the teams who ranked in triple digits nationally in turnovers lost last year: Louisville, Baylor, and Clemson. I’m not saying you want to commit turnovers, but they are hardly the death knell for an offense.

LSU has lived in terror of turning the ball over since Jarrett Lee’s freshman year. It has absolutely stifled the offense. I don’t want the team to get all loosey-goosey on ball security, but double digit interceptions might reveal an offense that occasionally threw a risky pass. LSU has consistently done better at not turning it over than it has at scoring. Only once in the past six season has LSU ranked higher in PPG than TO Lost.

This extreme risk averse philosophy shows up in the stats in an even more obvious place: number of plays.

There’s been a lot of talk this offseason that the way to beat Alabama is by sheer number of plays. While I think that’s overly reductive, LSU’s slow pace certainly isn’t helping.

Efficiency leads to explosiveness precisely because it gives you more bites at the apple. LSU earns those extra bites, but decides to instead throw the apple in the trash. Only in 2013 did LSU’s great YPP numbers lead to a large number of big plays. It’s like our shooter is being left open beyond the arc, and we don’t take the shot.

LSU was never a fast-break team, but the pace in recent years has become positively glacial. The past two seasons, LSU has been one of the slowest teams in the nation, taking over 200 fewer offensive snaps than the average team.

Now, there can be a good reason to play slow. If you have a great defense and lousy offense, it can pay to try and shorten the game. While LSU’s offense hasn’t been the model of production, it has the component parts to be productive. Hell, the component parts HAVE been productive. The only thing holding the unit back is the lack of play calling, literally.

Another reason to call fewer plays is if you are talent deficient. The shorter the game and the fewer number of plays, the more likely the outcome is determined by bizarre variance or dumb luck. Bad teams want to keep it close and hope the ball bounces their way. You do that by shortening the game, and having less chances for talent to win out. LSU plays like an underdog, and by doing so, they increase the chances of getting upset and underperforming their talent levels.

LSU has run the offense over the past decade as if they are a bad team. Even in 2013, LSU ranked 90th in the nation in total offensive snaps. NINTIETH. We, by design, took the ball out of the hands of Hill, Mett, Beckham, and Landry. That’s insane.

LSU has had a top-tier offense on its campus the past two seasons, and the only reason it didn’t convert that production into points is that we voluntarily slowed down the pace. The offense turtled for two seasons.

Matt Canada doesn’t need to build an offense. He just has to play the one we already have. Run more plays, the big plays will come. With big plays will come more points and with more points, more wins. The numbers don’t lie, even when it seems like they are.