clock menu more-arrow no yes

Filed under:

Fun with Charts: Returning Starters, Experience, and the Glory of X- and Y-Axes

New, 7 comments

Mississippi State lost its starters, and therefore it is going to lose lots of games. But what does it actually mean to return starters to a football team? And what if the number of a team's returning starters alone had little if anything to do with how good a team is? These and other existential mysteries are examined within.

Mullen holding onto a chart during a game. Did I make it? Probably.
Mullen holding onto a chart during a game. Did I make it? Probably.
Stacy Revere/Getty Images

After establishing that Mississippi State’s football team isn’t going to amount to doodly squat this year, I felt compelled to know more. Yes, we are going to suck; but how much are we going to suck? Is there a way to quantify our anticipated level of suckitude? And if so, could I synthesize the quantum of suckiness in a clear and straightforward manner to present to the remnant of our fanbase that hasn’t yet run for the hills?

My search for knowledge of the suck kept bringing me back to a simple set of undeniable facts, presented with ample clarity by Phil Steele’s staff of statistical wizards:

Phil Steele - returning starters

State fans hither and yon have tried to explain these facts away, but they are inescapable. Compared to most of our SEC brethren, we just don’t have the players. Returning starters are the ones with the experience, and experienced players in college football are essential for success, especially in a division as hellacious as the SEC West.

That’s what I assume, anyway. I mean, with more starters coming back, you’ll have more developed players the next year, more depth, more playmakers, more room for error, more variety in your play-calling, and, ultimately, more wins.

Right?

The first major premise seems to be that a team’s number of returning starters represents the amount of experience the team brings back from the preceding year. The starters are presumably the best at their respective positions, so they’re the ones making all the tackles, interceptions, touchdowns, and so forth. In other words, when you lose starters, you lose your playmakers.

To explore this concept in more detail, I decided to look beyond returning starters and discover what each SEC team actually had coming back—how many touchdowns and yards of rushing, receiving, and passing on offense, and how many interceptions, pass-break-ups, sacks, tackles-for-loss, and tackles made by top-tacklers (those that had ten or more tackles last season) on defense. I then compiled all that information in a couple of fancy graphs that show the data for each team in the SEC in terms of percentages—the percentage of a team’s 2014 sacks, rushing yards, and so on made by a player returning for the 2015 season.

[Caveat: This only tracks returning experience, not the quality of the players or units. For example, just because Vanderbilt has tons of experience coming back all over the field doesn’t mean that their team takes a dramatic leap forward next season. Sure, Vanderbilt may be bringing back 90% of its sacks or rushing touchdowns, but they didn’t have many sacks or rushing touchdowns to begin with.]

Here are the numbers for the SEC’s defenses, listed in order of their final S&P+ rankings at the end of last season:

Defense Experience

Now, there’s a ton of stuff going on in that chart, so let me just focus on a couple of things. First, note how the top six defenses in the SEC last year on the whole seem to have similar amounts of experience coming back, with the possible exceptions of Alabama and Arkansas. Alabama is the only team among those top six to bring back 2/3 or more of playmakers in three of the five categories I listed. And Arkansas is the only one of the top six that brings back 50% or less of playmakers in three of the categories. Otherwise, those top teams all look relatively similar, especially compared to the bottom eight defenses. Even though the number of returning starters varies among those top six teams, they all seem have lost roughly the same amount of playmakers coming back.

Second, MSU’s defense, which lost a conference-high seven starters, does not seem any more lacking for experience than a number of its SEC-West peers that lost fewer starters. Look at Auburn, Mississippi, LSU, and Arkansas, for example. They each return two, three, or four more starters than State, but the gap between the percentages of their returning playmakers and State’s is generally small. This is especially the case compared to teams in the East that bring back two or more starters than State, like Tennessee, Kentucky, Florida, and Vanderbilt. Those teams all bring back far more playmakers on defense than most of the teams in the West.

Ok, so let’s take a look at the numbers for SEC offenses now, which are also listed in order of their final S&P+ rankings:

Offense experience

First, I acknowledge that this really only measures the amount of experience returning at the skill positions. I couldn’t think of a metric analogous to touchdowns or yards-gained by which to tally returning "playmakers" on teams’ offensive lines, so what we get is an admittedly limited view of things.

That said, the trend on this chart seems to vary a bit from the one about defense. The top two offenses in the conference, Alabama and Auburn, lost a ton of their playmakers, but after that, there really doesn’t seem to be too much correlation between offenses’ quality in 2014 and the amount of experience they’re bringing back. Some good offenses lost a lot, others didn’t. Some mediocre or bad offenses lost a lot, others didn’t.

Ok, so what do these charts tell us about returning starters and experience? Perhaps that they are related, but not so much as people emphasize in their preseason soothsaying. As for State specifically, I suggest you take a look at State’s place on the Phil Steele chart and then at the returning experience charts. Maybe (probably (absolutely)) I’m biased, but that doesn’t look like a team gutted by graduation and the draft.

If the correlation between starters and experience is somewhat tenuous, what about the bottom line? What's the relationship between returning starters and the quality of a team? Do teams with lots of returning starters play better and win more games than teams without as many returning starters?

Nope:

Wins and returning starters

The points on that handsome chart represent every Power 5 team in 2014, and the trendline shows the correlation of returning starters and wins. In a nutshell, teams with less returning starters tended to win more games last year than those with more.

And, just cuz I can, here’s a breakdown by conference, showing win averages of teams with 12 or less returning starters and of teams with 13 or more returning starters:

Wins and Returning Starters (summary)

The data behind the conference-specific chart is interesting. Only twenty-three Power 5 teams returned twelve or fewer starters for the 2014 season. (This amounted to four or five teams per conference.) Of those twenty-three, only three—Vanderbilt, Wake Forest, and Texas Tech—had losing records. And while only four teams returned nine or fewer starters like State will in 2015, their results were pretty solid—Oklahoma State won seven games, and Wisconsin, Baylor, and Missouri all won eleven.

Well, so, what the crap, man? Everyone’s harping about losing all these starters and how big of a deal it is, and last year teams with few returning starters won more games than teams with many? Without even getting to the merits of an argument for a causal connection, there's not even a correlation?

If that’s how it's going to be, then allow me to retort in kind.

Over the past ten years, thirty-one teams have had players return who had received Heisman votes the previous year. Here's a pie chart—who can resist a pie chart?—that shows those teams’ win totals for the years in which the Heisman-vote-recipient returned:

Heisman Pie

None had losing seasons. They each averaged about 10.5 wins and about 2.5 losses. Over 70% had ten or more wins. And about a dozen played in BCS-level bowls.

So, according to this overwhelmingly consistent and logically unassailable method of prediction, what does the SEC really look like this year?

This:

Dak chart