Football by the Numbers: Quarterbacks and YPA

We’re going to break down fantasy football here in the preseason By the Numbers. The idea is to develop models that will help us in our projections. We want to find the players who are overvalued and undervalued to lower our risk of drafting busts and increasing the odds of finding sleepers.

We start with the quarterback position. We first have to find the statistic, test its consistency in correlating to the fantasy categories we care most about and then apply it to the current group of signal callers. For example, we care most about projecting touchdown passes. But those plays are a small sample in a quarterback’s overall performance and thus can be greatly influenced by random factors. So we’re looking for a much bigger sample of his performance that also predicts TD passes.

And what can be bigger than every pass attempt? Simply divide passing yards by passing attempts. Over recent years, the elite (upper third) of the quarterback class has a YPA (yards divided by passing attempts) of 7.5 or more. The middle class QBs are 7.0 to 7.4. The dregs, 6.9 or less.

Let’s see what this has to do with TD passes.

In 2012, the 11 QBs who averaged 7.5 or more per attempt (minimum 25% of team snaps) threw 301 TD passes in 5,412 attempts, a TD rate of 5.6% or 27.8 TD passes per 500 attempts.

The eight QBs at 7.0 to 7.4 threw 187 scoring strikes in 4,244 attempts, 4.4%, 22 TDs per 500 attempts.

The 19 dregs below 6.9 tossed 248 TDs in 7,370 attempts, 3.4%, 16.8 TDs/500 attempts.

Other seasons need to be similar for us to bet on the YPA stat when it comes to TDs.

2011 - 7.5 YPA or more QBs: 5.5% TDs and 27.5 TDs/500; 7.0 to 7.4: 4.2% and 21.1 TDs/500 and 6.9 and under: 3.5% and 17.6 TDs/500.

2010 - 7.5-plus: 5.5% and 27.2 TDs/500; 7.0 to 7.4: 4.4% 21.9 TDs/500 and 6.9 and less: 3.9% and 19.6 TDs/500.

Seems pretty tight to me. The averages are a 5.5% TD rate for the top group, a 4.3% for the middle group and a 3.6% rate for the bottom group. That translates to 27.5 TDs/500 attempts, 21.5/500 for the middle and 18/500 for the rest.

But this only matters if we can project YPA. Research shows that we need only 400 attempts to establish a true YPA level, good news for Russell Wilson believers (remember to throw in those playoff games, too). TD percentage doesn’t become predictable until 1,125 attempts.

Of course, we also use yards in our game and it’s very obvious that the higher YPA quarterbacks pile up more yards as long as they are not far below league average in attempts.

The other caveat here is that the running QBs tend to underperform their TD pass rate, but we don’t care because they are running in those touchdowns (Cam Newton being the prime example here).

Let’s use this YPA model now to make some recommendations. We’ll go back three years so we can have more absolute certainty over a QB’s true YPA ability.

Three QBs with a YPA better than 7.5 in their last three seasons are currently huge draft bargains. Matt Schaub (Yahoo! ADP of 129), Philip Rivers (128) and Ben Roethlisberger (107).

Schaub (7.7 YPA since 2010) is the 24th QB drafted. The argument is that he’s not going to get a lot of attempts for the Texans, given their defensive prowess and running game. But last year, he threw about as much as Aaron Rodgers. He did go over 4,000 yards, too. And he had a 500-yard passing day. But the caveat here is that Schaub for his career has, for reasons I do not understand, badly underachieved with a 4.3% TD rate when it should be about 5.5%. Obviously, the bias in favor of running near the goal line is a factor here. But the whole idea of using models in projections is to do your thinking when building the model and then stop thinking and do what it’s telling you.

Rivers (7.8 YPA) is the 23rd QB taken on average. Like Schaub, he’ll be sitting on the waiver wires in most 12-team leagues with short benches. But one of them will finish top 12 in fantasy football points among 2013 QBs. Rivers is a rock solid bet for 28 TDs on a bad team with no running game and probably a bad defense. How high can the interception tax be in these leagues? How worried can we be about a new offensive system?

Roethlisberger (7.8) is the 14th QB off the board. Don't even think about drafting a QB early when you can get Roethlisberger this cheaply with that elite YPA on a team in transition defensively and with no proven running game.

Now is the time where I make enemies. Just as YPA tells us who is underrated, it tells us who is overrated, too.

Should we pass on Matt Ryan (7.2), Matthew Stafford (7.1) and Luck (7.0)? Ryan has been trending up in YPA and was 7.7 last year, plus he has elite skill weapons. I think he's being overdrafted still (ADP: 46) but not nearly as egregiously as the three-year sample suggests.

Stafford has Megatron and it's a good thing because, without him, he’s probably Mark Sanchez. But who cares, he has him. Again, though, why am I taking Stafford (63 ADP) over Roethlisberger? Note also that I used YPA last year to scream that Stafford wouldn’t come close to a 40 TD season again and the predictive value of the statistic was strongly affirmed.

We’ve already addressed Luck’s volume but I note that he has no great skill player to carry him, never mind two like Ryan has. And he’s way over the 400 passes we need to reasonably predict his true YPA ability. I don’t want to bank on volume from a QB for whom I have to pay this high a price for the position. And I realize that Luck can greatly improve, but I'm not going to pay retail for improvement that I have not seen.

Fantasy Football video from Yahoo! Sports:

Related coverage on Yahoo! Sports:
• Darrelle Revis could make preseason debut on Aug. 24
• Former NFL lineman blasts Cowboys QB Tony Romo
• Eagles acquire a wide receiver from Chip Kelly’s past