Does it Correlate?

As part of the discussion of my post yesterday, many ‘truisms’ about football came up.

Readers suggested that we instinctively know that line play matters for a variety of reasons.  One of the suggestions was that better line play means more rushing yards, which helps the Colts play action and makes the offense better. While that certainly makes logical sense, I wanted to test to see if it was accurate.

I’ve taken a variety of statistics to see if any of them produce any correlation to positive outcomes for the Colts since 1999.  I certainly don’t think that offensive line play never matters to any team, but merely that I don’t know how much it affects the Colts’ ability to win games and score points. The Colts present a rare opportunity to watch one offensive system with the same quarterback, and relatively low skill player turnover for an entire decade.  My hypothesis is that offensive line play has become less important to winning in recent years, and that Manning (and theoretically other QBs of a similar ilk) can mask poor line play.  The first step is to see which of the stats have a correlation on positive outcomes. Note:  These sample sizes are small.  That’s the nature of the beast.

I tested the following categories against one another:

Measures of O line play:

Rushing Yards-A simple crude stat

Rushing Yards per Carry-A measure of quality which presumably would be affected by line play.

Sack Rate–The Football Outsiders adjusted sack rate number.

Rushing DVOA-Another Outsiders stat that measures the effectiveness of run plays

Adjusted Line Yards-Another Outsiders number that seeks to isolate how many rushing yards per carry can be attributed to the Oline.

Measures of offensive success:

Wins-This is the goal of football.  Does line play lead to wins?

Points-This is the goal of offense.  To score.  Note: I did not adjust season point totals for defense/special teams scores.  If someone cares to go through the game logs and do so, I’ll rerun the numbers.

Offensive DVOA-The Outsiders measure of the effectiveness of an offense

Points/drive-How many points did an offense score each time it touched the ball.

Drive success rate-the percentage of down series that results in a first down or a touchdown

Summary of Results:

Correlation is indicated by a number approaching 1.  A negative number implies an inverse correlation (one goes up, the other goes down).  .25 and under is low correlation.  .25-.75 is moderate correlation.  .75 and above indicates a strong correlation.

Wins:  Rushing yards (-.52), YPC (-.61) and Sack Rate (-.41) all had moderate negative correlations.  Two of those are counter intuitive.  It makes sense that as the sack rate drops, wins would increase.  It doesn’t make sense that as rushing yards and yards per carry drop, wins increase, but for the Colts this past decade there’s some connection there.

Offensive DVOA:  Sack rate had a negative correlation (-.25).  Rushing DVOA and Adjusted Line Yards had a  positive correlation (.7 and .79).  So according to this, the more effective the line blocked for the running back, the more efficient on a play by play basis the offense was.

Points:  Rushing yards (.5), YPC (.47), and Rushing DVOA (.5) had some correlation with the offense scoring more.  Adjusted line yards (.83) had a very high correlation with points over the past decade.

Points per drive:  A better stat than pure points (it factors out defense and special teams scores, though not field position), there was one negative correlation.  Sack rate (-.46). The two positive correlations were rushing DVOA (.56) and Adjusted line yards (sensing a trend?) at .76.

Drive success rate: Sack rate had a negative correlation (-.36), rushing DVOA and adjusted line yards had positive ones (.59 and .67)

So, if we want to isolate the factors that BEST produce positive results, I’d say we should start with Adjusted Line Yards.  Over the past decade, they’ve shown the strongest correlations to DVOA, Points, and Points per drive.  In general, YPC is useless.  Sack rate has limited value (which is fine, because it probably has more to do with Manning than the line).  Rushing DVOA has value too, but that’s to be expected because it measures many of the same things that measured by the other stats.  It has little correlation with wins.  It also reflects the running back and not just the line.

So, Adjusted Line Yards it is.  Since that’s the one stat that best tries to figure out the impact of the O line, that’s convenient.  The one problem is that it doesn’t correlate strongly with wins.  We’ll set that aside for a moment, assuming that part of the Colts’ uniqueness has been the ability to win roughly 12 games every year with a variety of strengths and weakness. It could be that no single factor would correlate all that well with wins for the Colts.

Part of my theory is that something has changed in football in the last couple of years to make line play less important.  I graphed out the Colts major indicators to see if there was some change in the way the numbers were interacting.

What I found was disturbing.  Adjusted line yards have indeed been dropping for the Colts steadily since 2004.  However, almost every major offensive indicator has as well.  In part, I’m wrong.  The Colts offense is not thriving despite a steadily deteriorating line.  The Colts offense has been steadily eroding a little bit every year since the ’04 season.  There’s one huge caveat to that assessment.  It looks like the Colts started to rebound offensively last year, and if not for the final game and a half against the Bills and Jets, there’s a good chance that stats like points per drive would have come back to prior levels.  Adjusted line yards, however, hit their lowest point since the 2002 season.  So it seems like maybe the Colts were fixing the offense even as the line got worse, but without being able to isolate stats for just the starters last year, I don’t think I can prove that.

In short, these are my conclusions:

• Prior to 2008, Adjusted Line Yards definitely correlated strongly with offensive success.  A good run blocking line led to good offensive numbers.
• In 2008 and 2009, the Colts ALY plummeted.  The Offense, by most every measurement, suffered.  The O was a little better in 2009 than 2008, and it’s possible it was significantly better (depending on how much the final two games upset the numbers).  ALY hit it’s lowest level since 2002 in 2009.
• The Colts’ win totals were relatively unaffected by the offensive decline.  Part of the reason is that the Colts were an incredible 8-0 in one score games. A readers suggested the bad line play hurts the Colts in close games. That is demonstrably false.  The Colts ALY was much lower in 2009 and 2008, but the team went a remarkable 15-1 in close games during that span.  The team’s best ALY was 2004 (they lead the league).  They went just 3-2 in close games.
  

  	Record in “one score” games	Blowout wins (14+)
  2009	8-0	4
  2008	7-1	3
  2007	5-3	8
  2006	8-3	3
  2005	5-0	8
  2004	3-2	7

So in the now, does line play matter?  This year should tell us.  Reason would hold that the Colts can’t keep winning close games at this rate.  15-1 is an incredible run, and would indicate quite a bit of luck.

So does line play matter?

• It probably matters to the offense. 
• It probably has no bearing on close games.
• It doesn’t seem to have any effect on winning games.  The offense has slowed down to the point where the Colts aren’t blowing teams out as often as they used to, but continue to win close games at a greater rate than ever.  The result is that you can’t see the effects in the standings.