BBD: What does Department of Analytics mean?

http://www.buffalobillsdraft.com/201...nalytics-mean/

Quote:

---

When Russ Brandon announced the formation of a Department of Analytics a couple of weeks ago it left many fans wondering what that meant and what possible applications there were to the Bills talent evaluation process. Well you can count us among those fans who were interested in the answers to those questions. We are not experts in analytics nor do we have any intention of trying to act like it, so we turned to people who we consider experts.

---

Expert opinion on what it could mean for the Bills.

Nice.

I don't see how analytics can be effective in the NFL.

Baseball is such a different game. There's a bigger emphasis on individual players. Guys get to go one on one with a pitcher, positions are neatly divided, roles are clearly defined.

Football is this giant mashup of 11 guys to form one cohesive unit. Duties often become blurred. Talent at one position often goes unrewarded because the unit as a whole is terrible.

That's just my opinion though. It could be wrong.

Quote:

---

Originally Posted by JoeMama

I don't see how analytics can be effective in the NFL.

Baseball is such a different game. There's a bigger emphasis on individual players. Guys get to go one on one with a pitcher, positions are neatly divided, roles are clearly defined.

Football is this giant mashup of 11 guys to form one cohesive unit. Duties often become blurred. Talent at one position often goes unrewarded because the unit as a whole is terrible.

That's just my opinion though. It could be wrong.

---

While I agree baseball is more individual I would not discount this in football either. If you can get 11 guys that individually play to certain tendencies and you build a team off those tendencies this guy is good her this here and so on and so forth you could build a team that hides the other players weakness. You can even do it in game planning week to week and breaking down the other team. As humans we all do certain things without even realizing it. If you can find something an OT does say 80% of the time on a pass play etc you can give a huge advantage to your DE.

I have long thought that the current way of scouting is to much like guess work. People can say this guy is physically gifted he does this right and that right. Then he gets to the NFL and busts. There is more to this than just tendencies but I am a firm believer that if scouts studied more than just physical attributes and stats. Scouting would be farther ahead. People say Tom Brady was taken late. Well if you breakdown his college career he was fantastic. Far to many variables that only some scouts take into account such as surrounding talent, system that the player is in etc but if you break down his tendencies in certain situations you can debunk all those things and figure out what a guy is really about.

Just saw this... from Nate Silver, the king if analytics:

Quote:

---

Is it correct to assume that sabermetrics will never work in football and basketball like they do in baseball? And if so, is that because baseball is much more of an individual sport, or are there other reasons as well?

(Edit: By an individual sport, I mean that for the most part it’s pitcher vs. batter, with anything happening after that only a result of the initial matchup. This is not like football, where even a simple five yard run only happens because of many moving parts, i.e. blocking, and thus makes it much harder to grade anyone on a completely individual level.)
— AllDaveAllDay


A.
Well, I guess I'd put it like this: statistical analysis may not get you as far in basketball* or (especially) football as it does in baseball. But it still probably gets you much further than in most industries.


A lot of NBA teams (especially the ones that win a lot) have become VERY sophisticated about their decision making. Basketball may be closer to the baseball than the football end of the spectrum, both in theory and practice.

---

taken from www.fivethirtyeight.com

Quote:

---

Originally Posted by JoeMama

I don't see how analytics can be effective in the NFL.

Baseball is such a different game. There's a bigger emphasis on individual players. Guys get to go one on one with a pitcher, positions are neatly divided, roles are clearly defined.

Football is this giant mashup of 11 guys to form one cohesive unit. Duties often become blurred. Talent at one position often goes unrewarded because the unit as a whole is terrible.

That's just my opinion though. It could be wrong.

---

Is this a claim that your real identity is Bill Polian?

Quote:

---

Originally Posted by JoeMama

I don't see how analytics can be effective in the NFL.

Baseball is such a different game. There's a bigger emphasis on individual players. Guys get to go one on one with a pitcher, positions are neatly divided, roles are clearly defined.

Football is this giant mashup of 11 guys to form one cohesive unit. Duties often become blurred. Talent at one position often goes unrewarded because the unit as a whole is terrible.

That's just my opinion though. It could be wrong.

---

It's not the same approach as baseball. Football analytics has been around for quite some time. "Analytics" is a very broad term and there are a million ways to implement them. Different data, different formulas, etc.

i don't think that analytics can't be applied. i just think they'll need to develop the data points more thoroughly and find correlations across a number of categories.

there are no stats for say how many times a LB gets a sack that the OL were double teaming one player. Then it becomes a matter of finding guys who demand double teams, etc.

The analytics depart will have to be cutting edge and essentially tracking their own data. i don't think you are going to cut the current stats and find a different approach in the analytics.

Information is power!

Analytics is not going to take you from a perennial loser to the superbowl. It can however, improve your right/wrong decision making percentage by a few percent. When you make thousands of personel decisions, play calls, gameplans, etc., over the year, a few percent adds up. While that might win you one game the first year, the compounding over the next few years will leave you with a more talented and effective football team. If a good analytics department could result in improvements that allow us to win just two extra game three years from now, isnt that still a big win? Two games is the difference between 9-7 just missing a wildcard, and 11-5 with a first round home game.

Football is a game of inches. So take every inch you can.

Quote:

---

Originally Posted by Bendium23

Information is power!

Analytics is not going to take you from a perennial loser to the superbowl. It can however, improve your right/wrong decision making percentage by a few percent. When you make thousands of personel decisions, play calls, gameplans, etc., over the year, a few percent adds up. While that might win you one game the first year, the compounding over the next few years will leave you with a more talented and effective football team. If a good analytics department could result in improvements that allow us to win just two extra game three years from now, isnt that still a big win? Two games is the difference between 9-7 just missing a wildcard, and 11-5 with a first round home game.

Football is a game of inches. So take every inch you can.

---

Get off the msg boards Russ.

Analytics in basic form is what Professional Accountants do. It's the study of gathered data, and finding potential trends among other things, in an effort to improve the function, and or make changes to a business model to optimize performance of an organization, or company, etc. It can work in the NFL if it's done properly. JMO

Moneyball /= Analytics, but still found this tweet interesting:

@adbrandt: In varying degrees, yes. RT @whitesteve34 Are there any current NFL front offices using the Moneyball philosophy?

I think that like a lot of statistical tools, it has its uses. I can see it being of value in a number of areas, and could ultimately provide enough information to inform roster decisions one way or another. Much will depend on the criteria utilized though.
I seriously doubt it will ever become a sole determining factor in any football decision, but I can see it gaining weight over time, depending on its own success/failure rate.
I could see it also taking some effort away from the scouting process, in respect of elimination some candidates whose combination of performance and measurables don't add up, i.e. the establishment of baseline performance criteria, for a given position.
Something like that could allow for a more in-depth look at lower round candidates, by getting rid of the chaff earlier, for example.
Personally speaking, I see it as just an additional evaluation tool, that could help with making the most informed decision you can.

Quote:

---

Originally Posted by JoeMama

I don't see how analytics can be effective in the NFL.

Baseball is such a different game. There's a bigger emphasis on individual players. Guys get to go one on one with a pitcher, positions are neatly divided, roles are clearly defined.

Football is this giant mashup of 11 guys to form one cohesive unit. Duties often become blurred. Talent at one position often goes unrewarded because the unit as a whole is terrible.

That's just my opinion though. It could be wrong.

---

To a point you are correct. The human element is still very important in football but I think it's good to understand trends even if they don't make the decision by themselves.

I wonder how the analytics would have handled Russel Wilson. Statistically he graded as the most efficient QB in the draft last year (making him the #1 player on the board potentially). Of course if the analytics plugged in his height and weight it would have probably delivered an undraftable grade as very few guys have made it at 5'11"

Seattle clearly only cared about the 1st sentence and his game tape.

I think eventually when all front offices inevitably start using the same systems that the best teams will be the ones who spit in the face of the analytics. Data says you need to spend money at Position X, Y, & Z? Fine, let the other 31 teams chase those guys and I'll collect the best players at positions A, B, & C. Winning in the NFL has often been a result of developing a strength that gives you a clear competitive advantage. Another way is to go against conventional wisdom at just the right time. After these spread offenses become the norm I think a return to the power run game will prove it's worth, though I suppose that may have already happened (or never happened depending on how you look at it) ---just look at the teams still in this race. See also: Alabama. And Stanford knocking off Oregon. NE has transformed itself into a run 1st team this season BTW

Speaking of the wide open attack, it's nothing new. Don Coryell laughs.

"Analytics" are only as good as the person(s) doing the analysis. It is impossible to reduce human performance to single numbers except for very linear and objective things, such as the things that are measured at the combine, but how often are those indicators deceptive.

While I enjoy both PFF and FO, neither encompasses a number of things, such as the level of [individual] competition as merely one example. Another is the impact of a particular system in which a player plays. Some players simply play well in alternate or certain systems.

Etc. Analysis is a tremendously open-ended thing. The key to it is being able to synthesize the information that is gathered and convert it to some useable form, which is often easier said than done for players other than say QBs, RBs, or WRs.

For our franchise, I am not under the impression for a moment that anyone on the staff has the first clue what to do with good data even if it were handed to them gift-wrapped. Who is the Ernie Adams on this team? There isn't one. IMO this is just a passing fad for a bunch of dopes at OBD.

Quote:

---

Originally Posted by The Compromise

"Analytics" are only as good as the person(s) doing the analysis. It is impossible to reduce human performance to single numbers except for very linear and objective things, such as the things that are measured at the combine, but how often are those indicators deceptive.

While I enjoy both PFF and FO, neither encompasses a number of things, such as the level of [individual] competition as merely one example. Another is the impact of a particular system in which a player plays. Some players simply play well in alternate or certain systems.

Etc. Analysis is a tremendously open-ended thing. The key to it is being able to synthesize the information that is gathered and convert it to some useable form, which is often easier said than done for players other than say QBs, RBs, or WRs.

For our franchise, I am not under the impression for a moment that anyone on the staff has the first clue what to do with good data even if it were handed to them gift-wrapped. Who is the Ernie Adams on this team? There isn't one. IMO this is just a passing fad for a bunch of dopes at OBD.

---

If you read the article PFF actually does look at scheme:

"As far as differentiating between two players, again I think the proper numbers can be very useful. There will never be one magic number that completely distinguishes one player from another, though the closest thing we have is overall PFF grade. It’s a good way to evaluate a player’s performance, though scheme, surrounding talent, and other factors always need to be considered within the context of the grades. But even beyond that, our film study breaks players down in all aspects of their game, and it does so against players who play the same position, so it’s a big advantage when trying to decipher a player’s strengths and weaknesses."

Now where I do agree with you is that any analysis can yield bad conclusions. It can come from bad data, collecting the wrong data, or from drawing the wrong conclusions. I work in R&D at a big company and I can tell you that very smart people draw bad conclusions all the time. Often, you have to experiment over and over just to figure out how to properly run the experiment to get the information you want. And you can have a room full of people looking at the same chart and have a room full of different conclusions.

I don't think we have enough metrics yet and that they can be refined. But, I do think it's possible to measure just about anything. The experiment is usually in figuring out how to measure what you want information from. QB decision making, pocket presence... QB metrics would be really interesting. I think it's possible but we may not be there yet.

It will ultimately come down to how the data is being used and what were trying to find.

Quote:

---

Originally Posted by X-Era

If you read the article PFF actually does look at scheme:

"As far as differentiating between two players, again I think the proper numbers can be very useful. There will never be one magic number that completely distinguishes one player from another, though the closest thing we have is overall PFF grade. It’s a good way to evaluate a player’s performance, though scheme, surrounding talent, and other factors always need to be considered within the context of the grades. But even beyond that, our film study breaks players down in all aspects of their game, and it does so against players who play the same position, so it’s a big advantage when trying to decipher a player’s strengths and weaknesses."

Now where I do agree with you is that any analysis can yield bad conclusions. It can come from bad data, collecting the wrong data, or from drawing the wrong conclusions. I work in R&D at a big company and I can tell you that very smart people draw bad conclusions all the time. Often, you have to experiment over and over just to figure out how to properly run the experiment to get the information you want. And you can have a room full of people looking at the same chart and have a room full of different conclusions.

I don't think we have enough metrics yet and that they can be refined. But, I do think it's possible to measure just about anything. The experiment is usually in figuring out how to measure what you want information from. QB decision making, pocket presence... QB metrics would be really interesting. I think it's possible but we may not be there yet.

It will ultimately come down to how the data is being used and what were trying to find.

---

I have followed both PFF and FO (FO for the last few years, PFF the last year), and you are correct. PFF does account for scheme. So does FO. Neither claims to be the "only and final anwser," but both make very good use of more effective data. Film study in a traditional sense will always have a place, but we are smarter now and this is a BIG business. EVERYTHING has to be accounted for and researched. All data can be misleading which is exactly why it should be sliced and diced to get a more real picture. At it's most basic level I can say I had 150 tackles last year and commentators will call me a "tackling machine" etc. The quality of the tackle is what's missed. Where did I make the tackle? How many yards were gained or lost before I made the tackle? Where did I start on the field compared to where the tackle was made? What did my first movement indicate assignment-wise? Did I fight through traffic? How long did that take? Did I shed 1 or more blocks? How long did that take? Was I taken high or low? Did I hit the ground and come back up? Was I held at all? Was I fooled on a fake or did I diagnose quickly? Etc, etc, etc. Question after question to get more and more data. I think it's great that they are catching up with the rest of the century.

j/w Do they account for travel and weather?

And of course it's still a human game. Denver was analyzed as the perfect team by Aaron Schatz, with Baltimore having too many flaws to overcome. Whoops!

Quote:

---

Originally Posted by Mouldsie

j/w Do they account for travel and weather?

And of course it's still a human game. Denver was analyzed as the perfect team by Aaron Schatz, with Baltimore having too many flaws to overcome. Whoops!

---

Good points re travel and weather. I dont believe so. at least I don't recall. You're right that it's still a human game. PFF and FO will admit to as much too. As I said, it's a tool to make more informed decisions. There is no fool proof antecdote. I missed on Denver too. Even Schatz who is as good as they come doesn't win them all. At the end of the day you want all tools available to make the best decisions.