Some More Analysis of Steelers Takeaways and Turnovers

This has turned out to be a continuation of yesterday's post Take It Away, Steelers!

A couple of the people who commented on that post suggested looking at some other aspects of the data I collected. One suggestion was to calculate the point swing that an offensive turnover or defensive takeaway represents. This sounded like an interesting idea to me. With the help of the graph found on Advanced NFL Stats I figured out how many points the field position at the start and end of the play theoretically represented, both for the team gaining the ball and the team losing the ball. This gives us the point swing.

The classic example of this was given in the first comment, by One4theotherthumb—James Harrison's interception/TD in Super Bowl 43.  Harrison's TD not only put 7 points on the board, it almost certainly prevented an ARI TD, as he intercepted it in the end zone, and therefore represents a 14 point swing.

So here's what I came up with. Please note that I'm assuming a TD is 7 points, since hardly anyone ever misses the extra point.  Offensive TOs will be represented by a negative number, since they are the sum of the points the offense theoretically should have had added to the points the opposing offense theoretically are likely to gain. Defensive TAs are, naturally, a positive number, as they represent the points the opposing offense theoretically stood to gain added to the points the Steelers offense theoretically should have managed. The actual result of the turnover/takeaway is in brackets at the end of the line. The "Total Point Swing" is naturally from the Steelers' point of view.

Game 1 2010

TO 1: Lost at ATL 35 (2.9) + gained at ATL 28 (.9) = -3.8 [ATL FG]

TA 1: Lost at ATL 21 (.3) + gained at ATL 30 (3.2) = 3.5 [0-missed PIT FG]

Total Point Swing: -.3

Final Score: Steelers 15  Falcons 9

 

 

Game 1 2011

TO 1: PIT 44 (1.6) + PIT 37 (2.7) = -4.3 [BAL TD]

TO 2: PIT 24 (.6) + PIT 18 (3.9) = -4.5 [TD + 2 pt. conversion]

TO 3: PIT 20 (.3) + PIT 17 (3.9) = -4.1 [0]

TO 4: BAL 18 (3.9) + BAL 10 (-.3) = -3.6 [FG]

TO 5: BAL 20 (3.8) + BAL 21 (.3) = -4.1 [0]

TO 6: BAL 39 (2.6) + BAL 14 (0) = -2.6 [0]

TO 7: PIT 20 (.3) + PIT 9 (4.1) = -4.4 [FG]

Total Point Swing: -27.6

Final Score: Steelers 7 Ravens 35

 

Game 2 2010

TO 1: TEN 42 (2.5) + PIT 37 (2.8) = -5.3 [TEN FG]

TA 1: TEN 34 (1.1) + ATL 46 (1.7) = 2.8  [0-PIT turned the ball over to TEN]

TA 2: PIT 16(4) + PIT 20 (.3) = 4.3 [PIT FG]

TA 3: PIT 42 (2.5) + PIT 45 (1.6) = 4.1 [0]

TA 4: TEN 23 (.5) + TEN 23 (3.6) = 4.1 [0]

TA 5: TEN 42 (1.4) + TEN 35 (2.9) = 4.3 [FG]

TA 6: TEN 29 (1) + PIT 46 (1.5) = 2.5 [FG]

TA 7: TEN 20 (.3) + TEN 15 (3.95) = 4.25 [FG]

Total Point Swing: 21.05

Final Score: Steelers 19 Titans 11

 

Game 2 2011 - no TOs or TAs

Final Score: Steelers 24 Seahawks 0

 

Game 3 2010

TO 1: PIT 17 (.1) + PIT 31 (3.2) = -3.3 [TB FG]

TO 2: TB 32 (3.1) + TB 32 (1.1) = -4.2 [TD]

TA 1: TB 35 (1.15) + TB 44 (2.3) = 3.45 [PIT TD]

TA 2: PIT 17 (3.9) + TD = 10.9

Total Point Swing: 6.85

Final Score: Steelers 38  Buccaneers 13

 

 

Game 3 2011

TO 1: IND 47 (2.1) + MID (1.9) = -4 [IND FG]

TO 2. IND 42 (2.5) + TD = -9.5

TO 3: PIT 20 (.3) + PIT 12 (4) = -4.3 [FG]

TA 1: IND 25 (.6) + TD = 7.6

Total Point Swing: -10.2

Final Score: Steelers 23  Colts 20

 

Game 4 2010

TO 1: PIT 15 (0) + PIT 36 (2.9) = -2.9 [18 seconds left in game]

TA 1: BAL 22 (.4) + BAL 27 (3.3) = 3.7 [0-missed PIT FG]

TA 2: BA 17 (.1) + BAL 33 (3) = 3.1 [0-missed FG]

Total Point Swing: 3.9

Final Score: Steelers 14  Ravens 17

 

 

Game 4 2011

TO 1: PIT 40 (1.3) + HOU 11 (-.3) = -1 (10 seconds left in game)

Total Point Swing: -1

Final Score: Steelers 10 Texans 17

 

The additional question posed by steelbit was slightly confusing to me, but I think s/he was asking whose fault the resulting points are on an offensive TO. This makes some sense if you think of a TO deep in enemy territory as equivalent to a good punt.  Of course, you also give up the fairly high probability you had of scoring points on that drive. But the defense ought to be able to make a stop on a low-probability drive for the opposing team. On the other hand, if the offense gives the ball to the opposition within field goal range, it's hard to blame the resulting field goal on the defense.

So let's look again at those results and see how the defense fared after takeaways. The theory is that if the opposing team gets the ball at a "probability point" (henceforth PP) well below 3, then in theory the defense ought to be able to hold them to no points. If the opposing team gets the ball at a PP at or near 3, the defense can't be blamed for giving up a FG, although we might not like it. If the PP is considerably higher than 3, then the defense should be commended for holding them to a FG, and so on.  In other words, if the defense performs better than the probabilities would suggest, they are playing well.

By the same token, if the offense gets the ball back, courtesy of the defense, at a high PP, they "owe" it to the defense to convert it to some points. If they get it at a low PP, they can't really be blamed for not converting it to points, although the longer they stay on the field the better off the defense will be on the next series. And again, if the offense is converting more points than expected, they are performing well, and of course the converse is true.

Both of these are a completely different question than whether the offense/defense are generating greater/less than expected TOs or TAs. I briefly attempted to address that, and added for your viewing pleasure kk99's comparison chart. 

So here goes.

In Game 1 of 2010 the Falcons got the ball at a PP of .9, and converted it to a FG. The Steelers defense can be blamed for this, since we aren't considering the question of whether the offense should have given it up in the first place. Conversely, the Steelers offense was given the ball by the defense at a PP of 3.2. They did manage a FG attempt, which failed, so in this case the blame is put squarely on one person, as one could consider that the offense did their job.

Game 1 (2011) was, of course, a debacle. But how did the defense do with the admittedly seriously crappy hand they were dealt? These scores could be placed at the feet of the defense:

TA 1: PP 2.7, TD

TA 2: PP 3.9, TD + 2 pt. conversion

TA 4: PP -.3, FG

These points can't really be blamed on the defense:

TA 7: PP 4.1, FG

The remaining TOs were neutralized by the defense, but only TA 3 was a high PP TO:

TA 3: PP 3.9, 0 points

TA 5: PP .3, 0 points

TA 6: PP 0, 0 points

 

Game 2 2010

The one TO was at a high enough PP that the defense shouldn't really be blamed for the outcome:

TO 1: PP 2.8, FG

How did the offense fare with all of their gifts from the defense?

D'oh:

TA 1: PP 1.7, 0 [PIT TO]

TA 4: 3.6, 0 points

The offense couldn't really be faulted for not converting this one:

TA 3: PP 1.6, 0 points

They did their job on these:

TA 5: PP 2.9, FG

TA 7: 3.95, FG

They did better than expected here:

TA 2: PP .3, FG

TA 6: 1.5 FG

Game 2 2011 - no TOs or TAs

 

Game 3 2010

The defense can't be blamed for this score:

TO 1: PP 3.2, FG

This is another story:

TO 2: 1.1, TD

The offense did better than expected:

TA 1: PP 2.3, TD

The offense didn't have to do anything at all:

TA 2: run back for TD

Game 3  2011

The defense did worse than expected:

TO 1: PP 1.9, FG

Can't blame the defense for these:

TO 2. run back for TD

TO 3: PP 4, FG

Offense didn't have to do anything:

TA 1: run back for TD

 

Game 4 2010

Immaterial because the opposition got the ball on the wrong side of the field with 18 seconds left in the game:

TO 1: PP 2.9, 0 points

Like the situation in the Atlanta game, in both instances the offense did their job, and in both instances no points resulted because Jeff Reed screwed up. Both were over 40 yard attempts, admittedly, with the first being 49:

TA 1: PP 3.3, 0 points

TA 2: PP 3, 0 points

Game 4 2011

Again, there were only 10 seconds left in the game when Houston got the ball:

TO 1: PP -.3, O points

 

And for ZacP and Steelzombie, who disagreed as to whether luck was a significant factor in TOs and TAs, I did a bit of looking around to see what I could find on the subject of luck in general in football.  I definitely take Steelzombie's point that a disruptive defense is going to be more likely to get a TA, but on the other hand the football is shaped the way it is for a reason. Sometimes it bounces your way, sometimes it doesn't.

So what I found was a series of articles, also on Advanced NFL Stats, attempting to assess the role of luck in NFL games in general. (The link is for the final article of three, giving the gist of it as well as the payoff.) The writer came to the rather startling conclusion, which I thought he supported well, that about 52.5% of the results of games are attributable to luck. So if it is reasonable to extrapolate that to the TO/TA question, you're both right, sort of, but ZacP is slightly more right than Steelzombie. Feel free to read the articles and then hash it out, preferably in somebody else's comment thread : )

If any of you have any other ideas for stuff to look into, feel free to do it yourself : )  Here's kk99's elegant chart again. If you click on it you can see a bigger version that you can actually read, even if you're old like me.  Thanks again, kk!