Talk about the NFL

[ponchi101]

So far this week:
The Jets are NOT that terrible; they are not a gimme anymore.
The Giants are not for REAL, but they have to be taken seriously.
The Rams are pretty bad. They won, but that was not a dominant performance.
Seattle has scorched me every week. Have to start believing.

[ti-amie]

[ptmcmahon]

We do have it in TSN in Canada, but I might have been better off if we didn't have it

[ptmcmahon]

Well finally got a game where I can say (conclusively) going with the stats worked. 4th and 1 with 30 seconds left and Devner up four. Instead of kicking a field goal to go up seven and give Jacksonville the ball back, they go for it and convert. Jacksonville don't get another chance and game over.

So when it works as you can ... it's not a big deal No one talks about it when going with stats work.

if they hadn't converted and Jacksonville come back and win we would have heard about it for sure though.

[ponchi101]

Ok. Indeed, it worked.
But remember that part of my whining is when they go EARLY in the game and go for it in (as per your example) 4 & 1, and miss the three points and do not get them. Yesterday we saw one more example: the Packers went for it on 4&1, not cashing in the FG, and did not make it.
Of little consequence because the final gap was 10 points, but my question has been: in how many games has a team gone for it, not made it, and lose the game by a FG?
Which it is not what you are talking about, and indeed I may be just giving a good example of confirmation bias.

[ptmcmahon]

I just think it's way to hard to take a play that early in the game (think that play was in the 1st quarter or early second?) and say if they score three points there the rest of the game stays the same.

For example they would have played the last drive completely differently if they were down 7 than they did since they were down 10.

And that's not counting everything that could have changed in between as well. For example the next drive Buffalo wouldn't be taking over even at the same point in the field.

Although I'm sure something somewhere could calculate the % odds of winning based on A) Going for it and making it B) Going for it and missing it C) Field Goal attempt made and D) Field goal attempt missed. Some box scores have % chances of winning updated every play.

[ponchi101]

Agree. You play depending on the score. As you say, down by 7 is different than down by 10. It would have been a different game.
But that is sort of my point. Obviously, I don't know the depth of the analytics in the NFL. I am sure, though, that the basic stats are covered, FOR EACH PLAY. For example, every coach knows what are the stats for a given situational play: If you are at yard X, and you have 4 and so much to go, what are your statistical chances of converting by pass or rush, and what are the chances that your opponent will stop you, based on your and their previous performances.
What I am not sure are the long term models, i.e. models that predict the outcome of the game. Very early in a game, going for it on 4&1 as opposed to kicking the FG will obviously have some sort of chaos theory/butterfly effect modelling; you still have 3 more quarters to play. Those models collapse into more predictability the deeper you are in the game. Easy example: if you have 15 seconds to go, down by 4, and have 4&1 at the 10, the FG is an easy choice; it just won't work, so that option is discarded.
Those are the stats I wonder if they have.

[ptmcmahon]

Yeah, it's hard to know exactly how they are coming up with the win %. Are they just pulling from similar games and similar situations? Are they simming every possible play? Remember the famous Indy - New England game from years ago where New England went for it deep in their own territory. How do you come up with %s for that? You obviously can't compare every other situation where a team went for it fourth down in their own territory in the last minute..it may have never happened before. And certainly not with Brady and Manning as the QBs.

And every model seems to be different than others... like the one I quoted from a game a couple weeks ago where three "predictors" each gave a play outcome three different win % - and not even the same team for all three.

Basically there is just way too many variables to count on the % as being more than an educated guess most of the time - other than when a play makes it 100%

But it's fun to debate.

[ponchi101]

Exactly.
Another thing that I wonder at times.
You go for a very long FG; the kicker will have a certain % of making it. Of course, different kickers will have different percentages. The Ravens, with Tucker, are looking at a different set of data than Cano, also a fine kicker but not Tucker.
But.
You are exchanging a possible 3 for, also, a possible 7 for the other team, because you have to consider their chance of scoring WHEN given a favorable field position. For example, right now you do NOT give a favorable field position to the Chiefs or the Bills; it is too much of a gamble. So, if you go for it, your 3 becomes a (3 x %chance of making it) opposed to their (7 x %chance of scoring).
Early in the game, the probabilities are too hard to calculate, I think. So, go for the sure thing; that early in the game, the bird in hand beats those in the bush.

[ti-amie]

[ponchi101]

He has that look in the eyes of somebody that had a concussion 5 MINUTES AGO.
"Doctor: tell me where you are, what you do and your name.
Favre: Broadway, actress, Shirley Temple!"

[ptmcmahon]

Score another one for the stats people. After scoring a touchdown to get it to 24-23 with 18 seconds left, the Chargers decide to go for two. They convert and win 25-24.

I didn't see the win % but I have to imagine going to overtime is probably about a 50-50 chance... and going for two would have been a higher % chance.

Oops... and checking another box score Jacksonville did the exact same thing. Even scoring the touchdown to make it a one point game with 18 seconds left too!

[ponchi101]

Indeed, both LAC and JAX went for it on the last play, and made it.
I was watching the LAC game and LAC had a 77% success rate on 2-PT conversions, FROM LAST YEAR. It seems this was the first time they went for it this year.
Here, I felt that the stats were more complicated. If they went for the PAT, they would force OT. And then, a very complicated calculation would ensue.
Obviously, 50% chance of getting the ball back. Times your chance of scoring a TD. Versus 50% for the other team, times their chance for a TD. On and on. But this would have to be more than your 77% on 2PT conversions, so it seemed that, for the Chargers, this was the proper call.

[ptmcmahon]

Seeing 77% makes me think that means they were probably 7 for 9 last year. Another issue with these stats is they are such small sample sizes... hard to draw from 9 how true it is Since it was from last year would be different players, different opponents etc. Makes me really wonder when they do determine a % (not from real scenarios) how they do it.

So very hard to determine real numbers of course. I'd have to guess for both Chargers and Jacksonville kicking the extra point and going for over time would be somewhere near 50% - actually maybe even lower since extra points are no guarantee now.

Question then is what is really their chances of converting two points? I believe league average is above 50% so it's probably the right play. I imagine the Chargers would have had a slightly better chance of converting than Jacksonville. But then again, Jacksonville probably actually would've had slightly less than 50% chance to beat Baltimore - they were clearly the underdogs - so it may have even been a better decision for them to go for two as an "underdog."

Either way - yes complicated is definitely a good word for it!

[ponchi101]

The 49'ers, which were looking really good this season, lose Jimmy G for the rest of the season.
The man drafted last, Mr Irrelevant, has become Mr Relevant, in a hurry.