The Case for Receiver or Defensive Back, Part I

I’ve been rounding up a lot of Ravens’ statistics lately and crunching the numbers in various ways. One reason for doing this is to try to find out what statistics correlate better with winning and which of our past teams were strong in these stats.

Last week I wrote an article evaluating the best Ravens offense in team history using 28 different stats, which suggested that the top three offenses, unequivocally, were 1996, 2009, and 2012, depending on what variation I use.

The problem with that method, although it’s an interesting one, is that it only evaluates production in a vacuum – it doesn’t tell us which team should win more because each of these 28 stats correlates very differently to winning.

Fortunately, not being a statistician myself, (I don’t think one master’s level Stats class counts), some really smart people out there have done the work to find out what those are.

So, I turned to Advanced NFL Stats, one of the well-known leaders in this arena, for help.

And what we find out is that, as expected, Points are #1 – ‘Points For’ and ‘Points Allowed’ are the best metrics that correlate to winning for offense and defense, respectively.

#2? Turnovers. This probably does not come as a shock to you so far. Score more points = win game. Win turnover battle = score more points = win game.

What about yards though? The NFL ranks teams based on their yards/game during the season. Unfortunately, it’s a bad measure – Yards are statistically meaningless for correlation to winning. This shouldn’t be a shock. Going 70 yards down the field to kick a field goal is not as valuable as going 50 yards and scoring a touchdown, and the inverse is true for defense.

I haven’t yet figured out why the NFL uses yards but they do. This is why for instance in 2012 you heard Steeler fans talking about "being the #1 defense" despite having a bad year, losing to teams like the Raiders, Titans, and Browns. One reason for underperforming what their fans thought was a good defensive season and what they thought should have been a playoff caliber team was an abysmal turnover ratio. Pittsburgh was just 25^th in that department.

Now, of course, there are some problems with these particular stats in terms of helping judge value of a position on draft day. They’re great for evaluating how well a total team performed for a season, but you can’t evaluate a defender on "points allowed" or an offensive playmaker "turnovers prevented". How do we use stats to tell us what positions we should draft for?

One way is to figure out in terms of specific passing stats and rushing stats which contributes to winning best.

That old phrase "it’s a passing league" it turns out is very accurate: passing yards/attempt (i.e. offensive pass efficiency and defensive pass efficiency) are two of the best stats correlating to wins, as you see on Burke’s article. As he explains, this is odd because yards by themselves correlate very poorly. But when you count them per attempt, they matter a lot.

Don’t believe me? Here’s a table of the Ravens historical offenses sorted by Offensive Pass Efficiency:

Year/Variable	Z O Pass Eff
1996	6.5
2006	6.3
2010	6.3
2009	6.3
2012	6.3
2008	6
2011	5.9
1997	5.9
2001	5.6
2002	5.5
1998	5.5
2013	5.4
2005	5.1
2000	5.1
2007	5.1
1999	5
2003	4.9
2004	4.6

Four of the top five pass offense teams were tied at 6.3 yards per attempt and just happened to be four of our strongest playoff teams. 1996 however had one big problem and it was called the other side of the ball. Other takeaways are that Flacco obviously made a huge difference from 2008 onward as the lone common thread in all those years, and McNair made a similarly big impact in 2006. The Boller years look worse every time I look into the stats. 2013 stands out as an obvious example of what happens when you lose your two best receivers and your O-Line goes to hell (and yet even that is better than the four Boller years).

And here’s the table for Defensive Pass Efficiency:

Year/Variable	Z D Pass Eff
1999	4.6
2003	4.9
2008	5.1
2001	5.1
2005	5.2
2006	5.3
2000	5.3
2011	5.4
2010	5.8
2004	5.8
2002	5.9
2009	6
2012	6.1
1997	6.1
1998	6.2
2013	6.2
2007	6.8
1996	7

See anything in common? The teams with superior pass efficiency on offense or defense were generally playoff teams. Note how 1996 had our best pass offense but had our worst pass defense. Our worst years line up almost exclusively with one thing: a down year from Terrell Suggs, or the non-presence of Ed Reed. In case you were wondering why our defense seemed to turn a corner in 1999 after a terrible 1998, we happened to make a fairly impactful draft pick that year: All-Pro Cornerback Chris McAlister.

So, if we rank them now based on the simple net passing efficiency (i.e. subtract DPE from OPE), in theory our best teams should rise to the top:

Year/Variable	Pass Efficiency
2006	1.0
2008	0.9
2001	0.5
2011	0.5
2010	0.5
1999	0.4
2009	0.3
2012	0.2
2003	0.0
2005	-0.1
1997	-0.2
2000	-0.2
2002	-0.4
1996	-0.5
1998	-0.7
2013	-0.8
2004	-1.2
2007	-1.7

Pretty telling isn’t it? All the Playoff teams are right up at the top. 2006 was easily one of our most dominant and complete teams in every way. You might notice that the 2000 Ravens are somewhat missing from the top half of each table. There are many good reasons for that as I’ll discuss in the future. But to put it simply, it has to do with having an all-time run defense and a preposterously good Forced Fumble rate.

In any event, if correlation to winning is what we value (and obviously we value that), then it would appear anything that directly improves our Passing Offense or Pass Defense should be highly sought after come Draft Day.

Ozzie did a good job shoring up the pass offense in Free Agency but obviously Daniels and Smith are short-term fixes which is why Receiver is very viable as an early pick. But the aging of Ed Reed significantly degraded our pass defense after 2011 in a big way. Of course, other positions cannot be ignored, and BPA does mean something, but when Ozzie says he wants an athletic playmaker in centerfield, he’s got a good reason for that.