Analytics post mortem

[ibeattetris]

To be fair, I'd throw out the TFG data. Too many 3-and-outs and turnovers to skew the data.
2023 vs. 2024 is interesting. I've heard Key say he wanted to shorten the game and reduce an opponent's opportunities to score as a strategy. Need more game data for 2024 to draw firm conclusions.

2014: 20.2
2015: 22.5
2016: 21
2017: 23.1
2018: 19.4

Even in the final flexbone years we were averaging well more. Clearly, it is only three data points so far this year, but the fact of the matter is we have possessed the ball well below normal, which is going to drive our variance up even more than normal.

 

[roadkill]

2014: 20.2
2015: 22.5
2016: 21
2017: 23.1
2018: 19.4

Even in the final flexbone years we were averaging well more. Clearly, it is only three data points so far this year, but the fact of the matter is we have possessed the ball well below normal, which is going to drive our variance up even more than normal.

Yeah, it's going to be interesting how the average pans out over the course of the season. Reducing the number of plays was intentional, although I doubt this much reduction was expected. If this trend holds we are going to see significant drops in average points scored as well.

I wonder how our averages compare across the P4? I wouldn't put too much stock in early season data since most teams have only played one P4 game at best.

 

[ibeattetris]

I wonder how our averages compare across the P4? I wouldn't put too much stock in early game data since most teams have only played one P4 game at best.

I plan on doing this across NCAA later tonight when I have time. It's easy for a single team in excel, but won't easily scale without me writing a script for it.

 

[alagold]

On stats--goin g in to this game you have to give Syr high Off numbers as they are a good passing team but the flat obvious stat that sums up the game is our rush total and RB avg.Which failed us.Which goes to our number of our punts and not scoring enough. Just like in a lot of games this yr probably --we flat have to out score the opponent depending on OFF not our DEF..

 

[CuseJacket]

 

[FlatsLander]

2014: 20.2
2015: 22.5
2016: 21
2017: 23.1
2018: 19.4

Even in the final flexbone years we were averaging well more. Clearly, it is only three data points so far this year, but the fact of the matter is we have possessed the ball well below normal, which is going to drive our variance up even more than normal.

I think the F$U game was the first 14 total possession game in like 20 years or something. It's a major outlier. Plus, in the CPJ years, we scored a lot of points, meaning we had more drives rather than running death marches as much as some might think.

 

[ChristoGT]

There are new rules on when the clock continues to run after first downs. That could be a factor. Are number of possessions down in general?

 

[ibeattetris]

There are new rules on when the clock continues to run after first downs. That could be a factor. Are number of possessions down in general?

I looked, but so far it seems that the number of possessions are *not* down across FBS compared to last or the previous years. Total number of plays seems to be down though.

One thing I want to do later is do a splits based on % of run plays and show how that has influenced total number of drives over time.

I do think some of the early game seasons affects total number of drives. The Alabama vs USF game had 33 drives. USF trailed most the game, threw the ball 35 times and only completed 15 passes. Of USF 17 drives, 9 were under 1 minute 30 seconds and their longest drive lasted 3:34 on the first drive of the game.

I will admit that my initial hypothesis was wrong, and I am interested to see how the rest of the season plays out.

 

[4shotB]

I wanted to ask this question and didn't want to start a thread so I figure here is as good a place as any.

1) Do pro and college coaches put much stock in PFF grades? (I know they still must evaluate their own film so they can inform their own coaching and teaching).
2) As a corollary to the above, is it possible that a player's performance is evaluated differently by a coach than the raw PFF score. I guess another way to ask this, is it possible for a coach to say a player performed well even though the PFF grade is a 50 (or vice versa)? Say a player is a DL who gets blown off the ball all game long but somehow manages to force 3 fumbles when he or she did in fact make a play?

Thanks for any and all replies. As always on the internet, feel free to use any data, facts, anecdotes, hearsay, opinions, innuendo or feelings that support your position. I'll hang up and listen.

 

[roadkill]

I wanted to ask this question and didn't want to start a thread so I figure here is as good a place as any.

1) Do pro and college coaches put much stock in PFF grades? (I know they still must evaluate their own film so they can inform their own coaching and teaching).
2) As a corollary to the above, is it possible that a player's performance is evaluated differently by a coach than the raw PFF score. I guess another way to ask this, is it possible for a coach to say a player performed well even though the PFF grade is a 50 (or vice versa)? Say a player is a DL who gets blown off the ball all game long but somehow manages to force 3 fumbles when he or she did in fact make a play?

Thanks for any and all replies. As always on the internet, feel free to use any data, facts, anecdotes, hearsay, opinions, innuendo or feelings that support your position. I'll hang up and listen.

I decided to comment because PFF is relatively new to me and I had to dive into its methodology recently to understand it better.

My guess on your first question is that it probably varies a bit depending on the coach. Many probably use it as one of several metrics. I know coaches have spoken about “grading” players after a game but haven’t explained exactly how. Here’s an article where Belichick is quoted: https://bostonsportsmedia.com/2014/06/04/can-pro-football-focus-stats-be-blindly-trusted/

As for your second question, see above.

I have some issues with PFF grades. It is a subjective measurement masquerading as an objective grade. It can suffer from small sample sizes for some players. It implies accuracy to a tenth of a point, when the actual uncertainty is likely much larger. That said, I believe it is probably accurate enough to be useful in the broader sense, such as “player A had a pretty good game, player B did not.” Of course, at the coach or paid analyst level, they already know that from watching the game.

 

[4shotB]

I decided to comment because PFF is relatively new to me and I had to dive into its methodology recently to understand it better.

My guess on your first question is that it probably varies a bit depending on the coach. Many probably use it as one of several metrics. I know coaches have spoken about “grading” players after a game but haven’t explained exactly how. Here’s an article where Belichick is quoted: https://bostonsportsmedia.com/2014/06/04/can-pro-football-focus-stats-be-blindly-trusted/

As for your second question, see above.

I have some issues with PFF grades. It is a subjective measurement masquerading as an objective grade. It can suffer from small sample sizes for some players. It implies accuracy to a tenth of a point, when the actual uncertainty is likely much larger. That said, I believe it is probably accurate enough to be useful in the broader sense, such as “player A had a pretty good game, player B did not.” Of course, at the coach or paid analyst level, they already know that from watching the game.

Much apprecited roadkill. I couldn't get your link to work but quickly found it via Google. I think Coach Belicek has the credentials to opine on this and I thoughts his comments were very insightful. Thanks!

 

[Vespidae]

I wanted to ask this question and didn't want to start a thread so I figure here is as good a place as any.

1) Do pro and college coaches put much stock in PFF grades? (I know they still must evaluate their own film so they can inform their own coaching and teaching).
2) As a corollary to the above, is it possible that a player's performance is evaluated differently by a coach than the raw PFF score. I guess another way to ask this, is it possible for a coach to say a player performed well even though the PFF grade is a 50 (or vice versa)? Say a player is a DL who gets blown off the ball all game long but somehow manages to force 3 fumbles when he or she did in fact make a play?

Thanks for any and all replies. As always on the internet, feel free to use any data, facts, anecdotes, hearsay, opinions, innuendo or feelings that support your position. I'll hang up and listen.

I don't whether a specific coach does or does not put much stock in PFF grades on individual players. I think it's a waste of time. First, the only thing that matters is whether or not the team (and its players) achieved the desired outcome. As Saban said (and as does Key), "The only thing we want is that each player execute his portion of the play exactly as intended." In other words, in terms of the play expected, did the player execute or did he make an error? Lombardi said something similar in that, when it's 3rd and short, we know for a fact that will we gain three.

Bear Bryant said the exact same thing 60 years ago. Play in a way to minimize errors.

Moreover, all of the criteria for PFF are subjective. And with 21 other players on the field, sometimes, an individual player's result is an outcome of multiple outcomes for others. Impossible to manage.

The same thing happens in Texas No Limit when a player's action influences the actions of the remaining players. All you can do is to play the hand with the highest probability action that you can and hope for the best. (The goal in TNL is not win the most money, but to achieve the highest Expected Value, which statistically, will win you the most money.)

You can achieve analysis paralysis if you watch too much. In a more real-world example, when I was running a global business, we canned the 90 metrics we were asked to provide each week and instead, replaced them with three. Set a 40-year profit record.

Simplicity is good.

 

[slugboy]

Much apprecited roadkill. I couldn't get your link to work but quickly found it via Google. I think Coach Belicek has the credentials to opine on this and I thoughts his comments were very insightful. Thanks!

I’m not sure if the coaches are always sure what got audibled to. The scouts have to do some guesswork.

I wouldn’t go binary between “subjective” and “quantitative”. Refs and fans debate what a catch is—the sport has gray areas.

A game film breakdown is high information. Some opinions are low information. PFF is somewhere in the middle. So is a QBR. I’ll trust Santucci’s evaluation over a PFF rating, but I don’t have Santucci’s evaluation.

 

[4shotB]

Moreover, all of the criteria for PFF are subjective. And with 21 other players on the field, sometimes, an individual player's result is an outcome of multiple outcomes for others. Impossible to manage.



Simplicity is good.

I don't know if you read the article linked by Belicek , but regarding your statement #1, this is basically what Coach B said when evaluating players from other teams on film. If you don't exactly know what the specific assignment is for the guy you are evaluating, you can not grade his performance or wtte.

Your statement #2 - yes. 1000%. I have listened to Warren Buffet on more than one occasion (as well as Jimmy Buffet) and he agrees with us. Simplicity is indeed good. My second career was as a teacher - I think yours is similar albeit at the college level. I always told my students if they could not explain the math in a way that someone 2 or 3 years behind them in grade level could understand them then they didn't really understand the concepts themselves. Regarding Jimmy Buffett, he took the simplest song lyrics in the world (stepped on a pop top, blew out my flip flop, etc.) and leveraged into a multi-million dollar enterprise. I am going out on limb here but I am reasonably sure he would agree with you, me and Warren!

 

[ibeattetris]

"The only thing we want is that each player execute his portion of the play exactly as intended."

I am about 99% certain that this is one of the major components PFF grades on.

PFF Player Grades

The PFF philosophy, our process and how to interpret PFF Player Grades.

www.pff.com

Moreover, all of the criteria for PFF are subjective. And with 21 other players on the field, sometimes, an individual player's result is an outcome of multiple outcomes for others. Impossible to manage.

A coach is also subjective. The main benefit of the coach is they know the play call and they know what the intended result is. It still sounds like PFF at least tries to get around this

Each position has its own grading rubric so our analysts know how to put a grade on the various expectations for a quarterback on a 10-yard pass beyond the sticks or what the range of grades might look like for a frontside offensive tackle down blocking on a “power” play.

Honestly, I really liked when Coach Johnson would answer how many missed assignments the team had after each game. It was probably as close as we are going to get to a coach providing the grading themselves, but until then, it's just always going to be some form of eye test. I do think it's at least interesting that PFF has access to the all 22 tapes. I want to know how average joe can get access.

I am actually really bearish on PFF overall, but I don't think the arguments laid out really combat what PFF tries to accomplish. My biggest issue with PFF is that even if they try their hardest, there is always going to be differences across reviewers and there is no way to fully rid oneself of internal biases. I assume PFF is directionally right, but I would never use it for anything more than supporting evidence.

 

[CuseJacket]

 

[roadkill]

Stats like some of Chad's that rank in terms of absolute numbers (like passing yards or first downs allowed) are distorted by the fact that GT has played more games than 95% of the other teams. We look better than we should in some, and worse in others. This should start to settle out after our first bye, but probably won't be even until after our second.

 

[Lil G]

Stats like some of Chad's that rank in terms of absolute numbers (like passing yards or first downs allowed) are distorted by the fact that GT has played more games than 95% of the other teams. We look better than we should in some, and worse in others. This should start to settle out after our first bye, but probably won't be even until after our second.

Can’t argue too much with a completion percentage though.
Haynes and the WR core are making more plays than not.

 

[GT33]

Can’t argue too much with a completion percentage though.
Haynes and the WR core are making more plays than not.

If we can just get King to get the ball out of his hands on time we'll have a lethal offense.

 

[stinger78]

If we can just get King to get the ball out of his hands on time we'll have a lethal offense.

May be a twofold thing: first, he may be consciously trying to hold down the INTs, and second perhaps knowing he can run gives himself extra “margin” to accomplish the first. Bottom line is we still cannot afford to give away possessions.