#UNCvsGT Postgame

[85Escape]

This is absolutely fantastic - great find and thanks for the share. I hadn't seen it before but am bookmarking it.

Agree. Really interesting. The fact that most of the "Big Plays" were by UNC (with three of them negative Big Plays) is interesting.

 

[85Escape]

And I love these stats for QB runs:

Player	Stat Line	Yards/Play	EPA/Play	EPA	SR	WPA
Sam Howell	8 carries, 61 yds, 1 TD, 3 Fum (3 lost)	7.62	-1.26	-10.07	62%	-27.0%
Jeff Sims	10 carries, 128 yds, 3 TD, 0 Fum (0 lost)	12.80	1.34	13.41	70%	15.0%

Bwahaha! I'm not even sure what WPA is but I'm guessing -27% isn't Heisman Candidate territory!

 

[4shotB]

Bwahaha! I'm not even sure what WPA is

I thought that was one of them there fancy beers that the hipsters drink like y'all were discussing in the "Boycott Sweetwater" thread.

 

[Enuratique]

This site is fantastic:

https://gameonpaper.com/cfb/game/401282641

I looked at the code that powers this site to understand how EPA is calculated, and it's (as best as I can tell) a machine learning model trained on historical data. But does anyone know how to understand the actual value? Is it the expected points for that offensive play? How do you interpret negative expected points?

 

[85Escape]

I thought that was one of them there fancy beers that the hipsters drink like y'all were discussing in the "Boycott Sweetwater" thread.

That could work! A new category, not India Pale Ale but instead Welsh Pale Ale

I like the site but I couldn't find a key to the metrics (like @Enuratique mentions above.) I guess that negative point plays are plays that are likely to provide points for the other team?

 

[Lotta Booze]

A decent explanation on EPA here

Expected Points Added, commonly referred to as EPA, is a measure of how well a team performs relative to expectation. For example, if a team starts a drive on the 50-yard line, its expected points to start the drive would be about 2.5. If the team ends the drive with a field goal, thus gaining 3 points, its EPA for that drive would be found by subtracting its expected points from how many points it actually gained, 3 – 2.5 or 0.5 EPA.

This same logic can be applied to individual plays. Say the Chiefs start with the ball first-and-10 from their own 25-yard line, where its expected points would be about 1.06. If Patrick Mahomes throws a 15-yard completion, making it first-and-10 on the KC 40-yard line, where the expected points is now 1.88, the EPA of that play would be 1.88 – 1.06 or 0.82. In other words, that completion increased the Chiefs’ expected points on that drive by just over three-fourths of a point. If instead of the completion, Mahomes was sacked for a 10-yard loss, making it second-and-20 from their own 15, the new expected points would be about -0.54 and the EPA for that play would be -0.54 – 1.06 or -1.6. This means the sack decreased the amount of points the Chiefs were expected to score on that drive by 1.6 points.

 

[85Escape]

A decent explanation on EPA here

Sweet! That's very helpful. Makes sense. Thanks!

 

[FlatsLander]

And I love these stats for QB runs:

Player	Stat Line	Yards/Play	EPA/Play	EPA	SR	WPA
Sam Howell	8 carries, 61 yds, 1 TD, 3 Fum (3 lost)	7.62	-1.26	-10.07	62%	-27.0%
Jeff Sims	10 carries, 128 yds, 3 TD, 0 Fum (0 lost)	12.80	1.34	13.41	70%	15.0%

Bwahaha! I'm not even sure what WPA is but I'm guessing -27% isn't Heisman Candidate territory!

If you hover over the WPA on the website, it says "Win Probability Added". So Howell's rushes made UNC 27% less likely to win.

 

[GTpdm]

And I love these stats for QB runs:

Player	Stat Line	Yards/Play	EPA/Play	EPA	SR	WPA
Sam Howell	8 carries, 61 yds, 1 TD, 3 Fum (3 lost)	7.62	-1.26	-10.07	62%	-27.0%
Jeff Sims	10 carries, 128 yds, 3 TD, 0 Fum (0 lost)	12.80	1.34	13.41	70%	15.0%

Bwahaha! I'm not even sure what WPA is but I'm guessing -27% isn't Heisman Candidate territory!

Wait...so EPA measures Expected Points Added. Howell's rushes actually subtracted 10 "expected" points, While Sims' rushes added 13 "expected" points. That right there accounts for the 23-point game differential, does it not?

 

[jgtengineer]

Wait...so EPA measures Expected Points Added. Howell's rushes actually subtracted 10 "expected" points, While Sims' rushes added 13 "expected" points. That right there accounts for the 23-point game differential, does it not?

Makes sense basically howell had 8 carries, 5 for 61 yards and a TD but 3 fumbles, so a little under 50 percent of his runs were turnovers, (we all know these were sacks but college counts those as rushes) because of that he was giving away points.

 

[LongforDodd]

Makes sense basically howell had 8 carries, 5 for 61 yards and a TD but 3 fumbles, so a little under 50 percent of his runs were turnovers, (we all know these were sacks but college counts those as rushes) because of that he was giving away points.

Is a college sack only inside the pocket or somewhere tackle-to-tackle?

 

[slugboy]

Is a college sack only inside the pocket or somewhere tackle-to-tackle?

Anywhere behind the line of scrimmage

 

[jgtengineer]

Is a college sack only inside the pocket or somewhere tackle-to-tackle?

Sacks are a defensive stat. They always count as rushing for the QB. The Statistician split them on the defensive side based on if it was an intended run or not. Its why we technically had 10 sacks, but 2 of them were credited as TFLS due to howell becoming an actual runner. In the NFL they count sacks as - team passing yards.

 

[jgtengineer]

Anywhere behind the line of scrimmage

Not exactly, its why we didn't give up a lot of sacks for options plays taht were blown up behind the line. They analyze intent of play as well in college.

 

[Heisman's Ghost]

Like many of us on this board, myself included, people look for any evidence that supports what they believe. Confirmation Bias is the most insidious type of bias, because more than almost anything we want to have been right all along. Our egos are fragile things.

Media types are no different and are likely worse than many on this board. The moment they see evidence that supports what they had been paid to predict they'll be all over it...whether that is us losing a game or UNC winning a game. "Howell puts the team on his back and rights the ship" is a story that has likely already been half-written. The media types had crowned Howell one of the big things this year, and any evidence to the contrary is seen as an outlier while the slightest glint of evidence to support it is touted over and over again by color commentators.

When Wake and NC State get through with him, Howell won't be able to put anything on his back besides his shirt.

 

[slugboy]

Not exactly, its why we didn't give up a lot of sacks for options plays taht were blown up behind the line. They analyze intent of play as well in college.

I don't think that running plays have ever been counted as sacks anywhere--it's pretty clear that sacks are just for pass plays. It's still a sack if it's a bootleg outside the tackle box, though.

 

[orientalnc]

Maybe this is what was said above. In college football a sack is lost rushing yards. In the NFL, a sack is subtracted from passing yards.

I believe sacks are included in the QB ratings, but I am not sure about that.

 

[jgtengineer]

Maybe this is what was said above. In college football a sack is lost rushing yards. In the NFL, a sack is subtracted from passing yards.

I believe sacks are included in the QB ratings, but I am not sure about that.

They are as the college QBR includes rushing. College also has a Passer Rating, which doesn't include sacks and only looks at completions, incompletions and interceptions. I was confused that he was just targeting a pass play where the QB gets tackled behind the line of scrimage, vs the QB simply being tackled behind the line of scrimmage. Were it gets tricky is draw plays. During the CPJ years we often had sacks not counted where the QB took off quickly but didn't make it back to the line of scrimmage because the stats guys thought it was a designed draw. We also had atleast a few "sacks" show up from option plays where the A back was a little to far forward (resulting in it being a pass)

 

[LongforDodd]

Sacks are a defensive stat. They always count as rushing for the QB. The Statistician split them on the defensive side based on if it was an intended run or not. Its why we technically had 10 sacks, but 2 of them were credited as TFLS due to howell becoming an actual runner. In the NFL they count sacks as - team passing yards.

what about a scramble? Anywhere behind the line a sack?

 

[jgtengineer]

what about a scramble? Anywhere behind the line a sack?

Yes, If it is clear the QB is scrambling to throw. Its usually pretty easy to tell you look at the way its being blocked. If its an actual draw receivers will run their routes then after a count of 3 or 4 usually start blocking. Then its just TFL or blown up draw. This is why sack numbers get corrected throughout the season. Same for why broken plays get counted as a TFL, same with a fumbled snap that results in a tackle or recovery.