The Game Ball Thread: Week 8 vs Chargers

[Marciano490]

Is it just me or does Rex have the least graceful running style of any recent Patriot? It works, but man, it's not pretty.

You don't find his gait elegant?

 

[SeoulSoxFan]

You don't find his gait elegant?

This is pretty good looking, I say.

@DeeepThreat
Rex Burkhead vs Chargers
7-7 for 100.0%, 68 yards, 0 TD, 0 INT, 9.7 YPA, 125.9 target rating

 

[Saints Rest]

I was about to post that it seems to me, without looking at any stats, that Rex gets about 9 yes every time he touches the ball. And there's the stats to back it up!

 

[InstaFace]

The announcers were Fouts and Eagle. Do you really think they even noticed Ghost going to the tunnel?
They didn't say a word about any of that.

If you know a better NFL announcer than Ian Eagle, I'd like to know who it is. Romo might get there as a Madden-esque color guy, but certainly for play-by-play he's the gold standard right now.

 

[Bellhorn]

The logic in that short article appears to me to be flawed.

Spoiler

A "typical" TD (i.e., not including situations where going for 2 is likely) is worth, in an absolute sense, 6 points plus whatever fraction of attempted PATs are made - so less than 7 but close to 7. According to that article it seems that the expected points after receiving a kickoff is 0.7. So actually scoring a touchdown and converting the PAT is worth almost 7 points, or almost +6.3 above expectation, sure, but not because the other team is expected to answer with 0.7. That's because you shouldn't deduct the 0.7 pts opponent's scoring expectation for the ensuing drive because then, for that logic's sake, you'd also have to add the expected scoring expectation from the average starting position of a team re-obtaining the ball after having kicked off. You'd also have to keep iterating on the following negative and positive expectations. It'd be silly.

The most sensible baseline is simply the expected points for a team after receiving a kickoff.

A safety should be valued at 2 pts, plus 1.6 pts (because in other situations the safety-scoring team wouldn't otherwise have reacquired the ball), minus whatever the scoring expectation of the team giving up the safety was at the average down/LOS of which safeties occur, minus the scoring expectation is of where the average position teams take over if the offense doesn't score when they are in that down/LOS (since that'd be the alternative to getting the ball at their own 40).

I'm guessing the differences between doing it correctly and doing it the way that article did it aren't worth the effort, and probably not even worth the effort I just put into this post, but I can't not.

OT, but this is a slow thread, so hopefully no one will mind. I don't think your objections are well founded here.

1) You're correct, of course, that the ultimate goal is to establish the value of a play/drive/whatever relative to the baseline of the expected value when it started.* But I don't think anyone disputes this. A necessary component of this process is establishing the actual value (in terms of expected points) of each scoring outcome, and that is what is being established here.

2) You don't need to keep iterating through all subsequent scoring possibilities, because the expected value of each successive term drops by approximately an order of magnitude each time. The 0.7 point value associated with the kickoff implies that the receiving team should be the next team to score approximately 55% of the time. This means that the second-order term would be roughly (0.55-0.45) * 0.7 = 0.07. And since, as you correctly observe, it becomes extremely messy to try to keep track of these subsequent terms, it is perfectly acceptable to just discard them, and truncate the analysis after the 0.7 point value associated with this kickoff. On the other hand, it wouldn't make sense to exclude this first term, because the effect is both certain and non-trivial.

* The play under discussion here is actually a particularly salient example of this, of course, since fielding a punt near the 11-yard line should hardly ever lead to an immediate safety.

 

[IdiotKicker]

You don't find his gait elegant?

No, but he does have calm eyes inside his facemask.

 

[SeoulSoxFan]

If you know a better NFL announcer than Ian Eagle, I'd like to know who it is. Romo might get there as a Madden-esque color guy, but certainly for play-by-play he's the gold standard right now.

Cannot agree more. I've posted before and will post it again. Ian Eagle is unfairly wasted by having Fouts next to him.

 

[joe dokes]

What happened near halftime with Gostolowski? He kicked a FG I think, and he along with the holder and other people were all standing around looking at the ground versus celebrating. He then kicked off and then he was shown running into the tunnel with his helmet off.
Did he maybe pull something or tweaked an injury because of bad footing?
Was at bar so couldn't hear sound.

It went un-addressed by the announcers. I thought someone lost a contact lens.

 

[Super Nomario]

A "typical" TD (i.e., not including situations where going for 2 is likely) is worth, in an absolute sense, 6 points plus whatever fraction of attempted PATs are made - so less than 7 but close to 7.

FWIW, that article is from before they moved the XP back, so at the time 7 points could pretty safely be assumed.

 

[Bowhemian]

If you know a better NFL announcer than Ian Eagle, I'd like to know who it is. Romo might get there as a Madden-esque color guy, but certainly for play-by-play he's the gold standard right now.

OK, I think Eagle is OK, better than most. But Fouts brings him down 5 or 6 notches. There were a couple point during the game where stuff was actually happening, and they were too busy yacking about completely irrelevant stuff to notice it. Ghost leaving the field was one of those times.

So I guess for Eagle it is guilt by association.

 

[Oppo]

Fouts provides less 'color' than most play by play guys. Pair Eagle with a good partner and he'd be widely accepted as a top guy.

 

[loshjott]

OK, I think Eagle is OK, better than most. But Fouts brings him down 5 or 6 notches. There were a couple point during the game where stuff was actually happening, and they were too busy yacking about completely irrelevant stuff to notice it. Ghost leaving the field was one of those times.

So I guess for Eagle it is guilt by association.

Isn't that a role for the sideline reporter, try to find out why a guy is going into the tunnel in the middle of a game? Either the sideline guy didn't find out so they didn't report it on air, or it was routine like a bathroom break so not newsworthy so they didn't report it. My guess is both.

 

[Bowhemian]

Isn't that a role for the sideline reporter, try to find out why a guy is going into the tunnel in the middle of a game? Either the sideline guy didn't find out so they didn't report it on air, or it was routine like a bathroom break so not newsworthy so they didn't report it. My guess is both.

Sure that is fair. But that wasn't the only time during the game that they were talking about something completely irrelevant. Meh, it doesn't matter, I still enjoy watching the games. I just dislike what Fouts has become. I used to enjoy his commentary. Not sure if he got worse, or if my patience got shorter as I got older. Or both.

 

[dbn]

OT, but this is a slow thread, so hopefully no one will mind. I don't think your objections are well founded here.

1) You're correct, of course, that the ultimate goal is to establish the value of a play/drive/whatever relative to the baseline of the expected value when it started.* But I don't think anyone disputes this. A necessary component of this process is establishing the actual value (in terms of expected points) of each scoring outcome, and that is what is being established here.

2) You don't need to keep iterating through all subsequent scoring possibilities, because the expected value of each successive term drops by approximately an order of magnitude each time. The 0.7 point value associated with the kickoff implies that the receiving team should be the next team to score approximately 55% of the time. This means that the second-order term would be roughly (0.55-0.45) * 0.7 = 0.07. And since, as you correctly observe, it becomes extremely messy to try to keep track of these subsequent terms, it is perfectly acceptable to just discard them, and truncate the analysis after the 0.7 point value associated with this kickoff. On the other hand, it wouldn't make sense to exclude this first term, because the effect is both certain and non-trivial.

* The play under discussion here is actually a particularly salient example of this, of course, since fielding a punt near the 11-yard line should hardly ever lead to an immediate safety.

I looked back at some of those articles and I don't 100% understand their expected-points models. Does anyone know where I can find data that does *not* try to include the expected points of the other team on the subsequent possession? Just "on average, how many points do teams score when starting a drive from here"?

edit: never mind.

 

[dbn]

Ok, I *think* I have it now. Below is a toy model. Tl;dr: I think a safety is indeed "worth" a bit less than 4 pts).

Spoiler

* Before the kickoff change, the average offensive drive (acquiring the ball by KO, punt, or TO) started at the ~27 yard line, and the average drive resulted in ~1.9 points. N.B. that this does *not* mean that the average drive that started at the 27 yard line resulted in an average of 1.9 points (I'd guess a bit fewer in points), but it's probably pretty decent if sloppy assumption.

* According to the linked-to advanced football analytics article, given their " expected points", which includes some adjustment for the other team's next possession, the difference between "expected points" with a 1st/10 on one's opponent's ~40 yd line compared to one own's ~27 yd line is ~1.9. So, we might expect a team starting a drive from the opponent's 40 to score on average ~1.9+1.9=3.8 pts.

Let's consider team "A" to have the ball on their own 2 yd line facing team "B"'s defense. Let's also call this possession "P1", the next possession (by either team A or B) "P2", etc. In this situation team A is expected to score, on average, very few points in P1 - let's say it's 0.1 pts. Let's consider two situations, "S1" where they do not give up a safety, and "S2" where they do give up a safety, and look at the expected points going forward. One last note: in S1 the ensuing drive for B, on average, would start beyond their own 27, but let's not bother trying to account for that.

S1 predicted average scoring by possession:
_____P1__P2__P3__P4__P5...
A____0.1_NA__1.9__NA__1.9...
B____NA_1.9__NA__1.9__NA...

S2 predicted average scoring by possession:
_____P1__P2__P3__P4__P5...
A____0.0_NA__1.9__NA__1.9
B____2.0_3.8__NA__1.9__NA...

So the only difference is in the first two possessions. In S1 B is ahead by 1.8 points, in S2 B is ahead by 5.8 points. Thus, in this simple toy model the safety gained them ~4.0 points. For reasons stated above, it's probably a bit less than that, so the previously-stated value of ~3.6 is probably pretty good.

 

[dbn]

One final thought on the value of a safety (... I can stop anytime I want, I swear). The safety is "worth" exactly 2.00000 points plus additional value from not letting the other team advance the ball (to either possibly score or inflicting worse field position after their eventual punt or turnover). Forcing a team to punt from, e.g., their own 2, gives about the same additional value. So, I now think the above post over-estimates the "worth" of a safety as most of that added value is due to backing the team up against their goal line to begin with.