Matt Cassel needs to throw MORE interceptions - by Jason Lisk

Getting hurt isn't ALL that bad.This is the second reader entry on UFR. Jonathan Tucker did a nice story on Greinke vs Sabes a couple weeks ago. This time Jason Lisk presents a compelling argument that Cassel needs to increase his odds of being intercepted because the benefits of taking that risk outweigh the disadvantages. My only added thought on this is that QB’s rarely get hurt throwing interceptions, but they sure do suffer physically for taking sacks. Anyway, this is a very good piece. Thanks Jason!

Matt Cassel needs to throw more interceptions.  And a few more incompletions wouldn't hurt either.  Absurd, right?  A Chiefs fan advocating that the quarterback needs to throw more, not less, interceptions, and bounce more passes into the turf.  I must be crazy.

This is, of course, partially a modest proposal.  I don't literally mean that he needs to stand up at the snap and throw the ball to the other team.  What I really mean is this: Matt Cassel and the Chiefs offense needs to stop taking so many sacks, and they need to adjust their behavior to take more risks in exchange for avoiding those sacks.  And these things are related.  An incompletion is a sack avoided.  An interception is a sack avoided.  And conversely, a sack is an incompletion or interception avoided.

Interceptions are bad, and we'll get to those in a bit.  First, why sacks are bad:

1) the yardage lost coupled with the loss of downs means first downs become far less likely; and

2) quarterbacks fumble on sacks at a reasonably high rate.

We've seen (1) in action.  I looked at the play by play record for the Chiefs' sacks this season.  The Chiefs have picked up 1 first down by an automatic face mask penalty on one of the sacks, and managed 2 other first downs on the set of downs in which the sack occurred.  Sometimes, multiple sacks have occurred during the same set of downs.  All told, 23 sets of downs have featured at least one sack, and the Chiefs ultimately punted 15 times, turned the ball over on downs 4 times, threw 1 interception, and made one field goal on fourth down following a third down sack.  That's one first down by penalty on the sack, and otherwise 2 for 22 in gaining another first down following the sack.  The Chiefs have had 84 offensive possessions (discounting kneel down situations) and almost a quarter of them have ended soon after a sack.

We haven't seen (2), yet, as Cassel has fumbled twice but the Chiefs recovered both.  The Chiefs currently rank first in the NFL in lost fumble rate per offensive drive, but that number will change if the sacks continue.

On Monday, Martin alluded to some research I have done on quarterback rate stats.  My research suggests that quarterbacks have a fair amount to do with determining the sack rate.  I've also seen every Chiefs game, so I know the Chiefs line has not played well.  I also recognize that Cassel doesn't have the best of circumstances in which to operate.  My own eyes, as well as measurements like rushing yards per carry, and negative plays allowed in the rushing game, or passing yards per attempt and completion percentage all suggest that the line has struggled. That said, I think that Cassel can eliminate some (not all) of the sacks by being willing to get rid of the ball when the protection breaks down.

There's an oft-repeated saying that "a sack isn't the worst thing in the world."  Taken literally of course, this is true.  There are a lot of things far worse than what happens during the game of football.  Even during the game, other outcomes may be worse than a sack. I hate that saying, though, because it is a generality that is not true in all circumstances.  In some situations, the choice to hold a ball and take a sack is the worst choice.

Let's talk some theory here.  If a quarterback (Quarterback A) throws 10 passes, completes 2 of them for only 20 yards, and throws 2 interceptions, I think we can all agree that is horrible.  A 20% completion rate, 2 yards per attempt, no touchdowns and 20% interception rate would easily garner the lowest possible quarterback rating.  If you watch game telecasts, you will often see them cite what a particularly quarterback does under pressure and when getting hit, and generally the numbers, even for top quarterbacks, are in the 40.0 to 60.0 quarterback rating range when being pressured or hit.  This is better than my hypothetical quarterback above, but still bad.

So I think we can agree that such a performance on 10 throws would be less than ideal.

But now let's assume that another quarterback (Quarterback B), instead of throwing those passes, holds the ball and takes the sack on all 10 attempts.  We'll give him 1 fumble lost on those 10 hits, and an average of 5 yards lost per sack.

Interceptions are costly, and the quarterback in the first scenario has more turnovers.  Without getting too deep into how to value an interception (because there has been a lot of research done on converting the value of an interception to yardage), let's just say for now that there is a lot of variation in interception cost (some are like deep punts, others are returned for touchdowns, and all results in between).  I'm comfortable, for this exercise, saying that an interception is a -50 yard play.  For simplicity, we'll also say that a fumble is a -50, but recognize that it has a lower variance (most are recovered near the line of scrimmage, but fewer are returned for touchdown).

------------------------------------

--For those interested, there is some great discussion from Brian Burke of Advanced NFL Stats and other commentators on measuring the value of an interception.

---------------------------------------

Using the round figure of -50 yards for turnovers, Quarterback A has generated -80 yards of value on 10 throws.  Quarterback B has generated -100 yards of value on the 10 sacks taken.  Quarterback A's worst results are worse than Quarterback B's (more turnovers, greater risk of defensive touchdown), but his best results are better than Quarterback B's, and his average result (an incompletion) is also better.

Which quarterback has cost his team more on those 10 plays?  You might be tempted to say Quarterback B based on the specific numbers I laid out, but I'm not really concerned with the specific estimates.  You want to quibble and change the fumble rate to only 1 out of 20 sacks when the quarterback decides to hold the ball, or change the interception rate to 25%, or some other tweak, that's fine.  I think the answer is this:  It depends on the situation.

Getting rid of the ball and taking incompletions or risking interceptions when you are not quite ready to throw, but have to, is the high variance play.  You want to get rid of the ball, rather than take the sack, when the cost of the turnover is less and when the situation demands risk taking.  Namely:

1) on third down;

2) when trailing; or

3) when you are the underdog and your team has a below average defense.

Holding the ball is the low variance play.  You want to hold the ball when the cost of turnover is higher and the situation demands risk avoidance.  Namely:

1) on first down (since you have two other downs in which to recover);

2) when leading; or

3) when you are the favorite and your team has an above average defense.

The problem is that most of the time Cassel has taken a sack, the Chiefs' situation (below average defense, the underdog, and already trailing when a lot of these sacks are occurring) mostly dictates a higher variance strategy.  Cassel is avoiding interceptions and near league average in that category, at the expense of taking too many sacks.

I'm going to close with taking a quick look at the "dumb" quarterbacks who avoided sacks at the risk of throwing too many interceptions, compared to the "smart" quarterbacks who avoided interceptions by taking too many sacks.  I used the pro-football-reference season finder to pull all seasons since 1978 where a quarterback threw the minimum qualifying pass attempts (224), and was average or better at avoiding interceptions, but took sacks at a rate at least 1.33 standard deviations worse than the average quarterback, and compared them to those who were average or better at avoiding sacks, but threw interceptions at a rate at least 1.33 standard deviations worse than the average quarterback.  We'll call them the interception throwers and the sack takers.

I then looked at these quarterbacks and the points per game scored by their teams in games they started, as well as their team's winning percentage in games started that season.  Here are the results.

IT – Interception Throwers; ST – Sack Takers

-

The results aren't particularly close.  The interception throwers did turn the ball over more, but the exchange was fewer punts, more points and a better winning percentage. Cassel's sack rate and interception rate is very much in line with the "sack taker" group.  Hypothetically, if we shifted 10 passes under pressure for Cassel instead of sacks, his sack rate would have fallen to a still above league average but acceptable 6.6%, and his interception rate would have climbed to 3.6%, his passer rating would have dropped, and most fans would talk about how much worse he is playing…

…And the Chiefs would have scored more points in the process, and maybe, just maybe, won one of those other games. Matt Cassel needs to be willing to throw more interceptions, particularly if they are a by product of fewer punts, more offensive plays, more yards, and fewer sacks.