Madden NFL 11 News Post

FBGRatings Seeks to Recalibrate Madden Ratings, Change Game

01:46 PM - January 24, 2011 by RaychelSnr

I recentely sat down for a talk with FBGRatings.com's Dan Berens to discuss his site's vision and what's going on over there today. The site is currently working on getting accurate ratings for every player using real hard data converted into the Madden ratings universe. Dan claims that when these numbers are plugged into the game, it plays much better and much closer to real life. Check out the interview below and also check out Dan's website to see what he's got going on!

Interview with Berens on the OS Radio Show on BlogTalkRadio

Game: Madden NFL 11Reader Score: 6/10 - Vote Now
Platform: PS3 / Xbox 360Votes for game: 96 - View All

Madden NFL 11 Videos

Member Comments

# 1201 michiganfan8620 @ 07/21/14 11:20 PM

Quote:

Originally Posted by DCEBB2001


	The "random" ratings you cite are based on correlative data. That means that there is a strong correlation between a graded trait and the probability that it matches to a particular value of an ungraded trait. You can guess what these are for certain positions. STR is based on the 1-rep maxes for the three core lifts: Bench press, squat, and clean. The three are graded separately and then averaged. Those averages are then re-graded. What you find is that many OL are actually not as strong as what EA has led you to believe, and many non-OL are actually stronger than their OL counterparts. Think about it this way. EA has a correlation of 93% between size and STR. That means, the bigger you are, the higher your STR rating. However, this isn't always true. Ever go to the gym and see that 400lb guy who can barely curl 25lbs? According to Madden his STR would be closer to 99 than to 1. How about the 175lb powerlifter who can bench press nearly three times his bodyweight? THAT is strong. As you can see, in reality, STR has far less correlation to weight than what Madden has erroneously trained you to believe. All of the players are rated using the same scales. Ray Rice is not stronger than an OT because he is a RB. He is stronger because HE IS STRONGER! I do nothing for the physical attributes based on position. All players, regardless of position, are rated using the same scale. If you would read our FAQ section, you would know this before posting. As for SPD, you are, once again, ERRONEOUSLY, rating players based on the 40 times. 40 times are important, yes, but mostly for the split times. Using the split times, we can create a graph of time v. distance to see how the players covered that distance. We can then differentiate said graph to determine the velocity and acceleration the player ran with during any point of his run. In this case between MJD and Charles, MJD had a better top-end speed (which is how EA rates speed when you look at their mechanics), but Charles had a better initial velocity (how EA grades the ACC attribute). Some players start fast (Charles) but have less top-end speed, while others start slower but have a higher top-end speed (like MJD). The data tells us this. If you only use 40 times, you must ignore ACC because the velocity will be averaged. By incorporating the split times, we can also solve for ACC, which Madden also rates.

Yes, but we are also a few years away from the last time these guys ran officially timed 40s. Charles is faster now, MJD doesn't look like the same explosive player he used to be. The STR thing, there are still some issues. A guy like Mike Martin, DL for the Titans can squat 700 lbs. Then you have Arian Foster, who can squat 500. 200 lb difference equals 1 point difference? Not to mention the fact that Madden is referring to on-field strength, not weight-lifting numbers. Weight numbers usually correlate, but there are cases where they don't. The 6'7 OL has to go lower while squatting, push higher while benching, etc than the 5'8 RB. Thus resulting in similar numbers, even though the 6'7 guy could be stronger.

# 1202 michiganfan8620 @ 07/21/14 11:27 PM

Quote:

Originally Posted by DCEBB2001


	His physical traits are the same. Since he is suspended, he isn't very useful. I am sure that once his suspension is lifted, he attributes will go back up to whatever the scouts deem is appropriate. Until then, this will deter anyone from plugging him in as-is.

Well, because he is suspended, then he is incapable of doing a finesse move? It just doesn't make sense if FBG is only going by his player traits. His production won't be there the first 4 games, but he still has those same traits. The only players using your ratings are sim players, who wouldn't plug him in anyway. And if they did, why would it matter? IMO it would be better to rate him the way he should be without a suspension (something that will only affect his production, not his traits), and allow it to be up to the player to use him.

# 1203 ggsimmonds @ 07/21/14 11:35 PM

Hey DCEBB I didn't want to get the rookie rating thread even more off topic so I will post in this thread.

I am assuming it does, does the operationalization of attributes fall under the terms of your NDA? It has to right?

That is the gist of my questions and concerns regarding your ratings.

# 1204 DCEBB2001 @ 07/22/14 12:16 AM

Quote:

Originally Posted by michiganfan8620


	Yes, but we are also a few years away from the last time these guys ran officially timed 40s. Charles is faster now, MJD doesn't look like the same explosive player he used to be. The STR thing, there are still some issues. A guy like Mike Martin, DL for the Titans can squat 700 lbs. Then you have Arian Foster, who can squat 500. 200 lb difference equals 1 point difference? Not to mention the fact that Madden is referring to on-field strength, not weight-lifting numbers. Weight numbers usually correlate, but there are cases where they don't. The 6'7 OL has to go lower while squatting, push higher while benching, etc than the 5'8 RB. Thus resulting in similar numbers, even though the 6'7 guy could be stronger.

The average player loses .005 on a 40 time for every year accrued in the NFL. I highly doubt that a RB, of all positions, is actually faster. I have Foster at 398 for the bench, and 625 on the squat with a clean of 430 compared to 350 for Martin.

Madden's use of on-field strength is wrong. Strength is strength. How you APPLY it is what matters. Some players are strong, but don't apply it on the field. That doesn't make them weaker, they just don't use it correctly.

PLEASE get it out of your head that Madden does everything right. Just because Madden uses an attribute one way does not mean they are using it correctly.

# 1205 DCEBB2001 @ 07/22/14 12:17 AM

Quote:

Originally Posted by michiganfan8620


	Well, because he is suspended, then he is incapable of doing a finesse move? It just doesn't make sense if FBG is only going by his player traits. His production won't be there the first 4 games, but he still has those same traits. The only players using your ratings are sim players, who wouldn't plug him in anyway. And if they did, why would it matter? IMO it would be better to rate him the way he should be without a suspension (something that will only affect his production, not his traits), and allow it to be up to the player to use him.

Sorry you disagree, but I go with what the grades say. I don't make this stuff up. I am an expert in data analysis. I interpolate and publish.

# 1206 DCEBB2001 @ 07/22/14 12:20 AM

Quote:

Originally Posted by ggsimmonds


	Hey DCEBB I didn't want to get the rookie rating thread even more off topic so I will post in this thread. I am assuming it does, does the operationalization of attributes fall under the terms of your NDA? It has to right? That is the gist of my questions and concerns regarding your ratings.

I can't talk about who provides the data, what team he works for, or give out the data in its primary form. What I can do is interpolate it freely and publish it under the guise of Madden ratings.

I choose not to discuss all of the details on how I rate players because I want sole ownership over how I do things. I have put in thousands of hours into this project since 2009, and to be honest, I want the credit for that work. I am willing to discuss theory, but not specifics.

Get what I am saying?

# 1207 ggsimmonds @ 07/22/14 12:28 AM

From the other thread

"So you are telling me that if a WR averages 20 YPC he should be rated very fast"

No that is not what I am saying. You are putting words in my mouth.

What I meant was take any player and look at their production. Then you break it down and determine how he arrived at that. In your 20 ypc you break that down further. Look at his yac. On passes thrown to him look at how far the ball traveled in the air. Consider all those things and extrapolate ratings. Of course this matters more for the mental or skill ratings as opposed to the physical ratings. It varies by rating though.
Then if you converted the numbers properly, and the underlying code is adequate, the player should be a reasonable recreation of his real life self.

"I would rather quantify the parts of each player, add them all up, and see what the outcome is"
But we know roughly what the outcome ought to be.

I stated earlier that I was confused and I think I know why -- we are actually probably in agreement; we are just using different terminology. what I am saying is physical attributes=traits, skills=production. Or something along those lines.
I know you probably cannot reveal how you operationalize something like route running. But could you give me a general idea? What do you look at? Is it the outcome of a play?

This post is already much longer than I anticipated but let me go back to the draft and video game. When scouting for the draft it makes sense not to use production because college ball is different from the NFL. So a WR got 1200 yards his junior year...means little. But grading NFL players based on their NFL production? Well that is better.

# 1208 DCEBB2001 @ 07/22/14 12:33 AM

Quote:

Originally Posted by ggsimmonds


	From the other thread "So you are telling me that if a WR averages 20 YPC he should be rated very fast" No that is not what I am saying. You are putting words in my mouth. What I meant was take any player and look at their production. Then you break it down and determine how he arrived at that. In your 20 ypc you break that down further. Look at his yac. On passes thrown to him look at how far the ball traveled in the air. Consider all those things and extrapolate ratings. Of course this matters more for the mental or skill ratings as opposed to the physical ratings. It varies by rating though. Then if you converted the numbers properly, and the underlying code is adequate, the player should be a reasonable recreation of his real life self. "I would rather quantify the parts of each player, add them all up, and see what the outcome is" But we know roughly what the outcome ought to be. I stated earlier that I was confused and I think I know why -- we are actually probably in agreement; we are just using different terminology. what I am saying is physical attributes=traits, skills=production. Or something along those lines. I know you probably cannot reveal how you operationalize something like route running. But could you give me a general idea? What do you look at? Is it the outcome of a play? This post is already much longer than I anticipated but let me go back to the draft and video game. When scouting for the draft it makes sense not to use production because college ball is different from the NFL. So a WR got 1200 yards his junior year...means little. But grading NFL players based on their NFL production? Well that is better.

Production never tells the whole story. Even great players can struggle to produce. Is Larry Fitzgerald a worse player now because he hasn't cracked 1000 receiving yards since 2011? That is why I don't trust stats. Stats don't tell me how fast, strong, agile a player is. They don't tell me how hard a QB throws a ball. That stuff is useless to me, and I don't/won't use it. I use the scouting data like every decent front office does.

# 1209 ggsimmonds @ 07/22/14 12:39 AM

Quote:

Originally Posted by DCEBB2001


	I can't talk about who provides the data, what team he works for, or give out the data in its primary form. What I can do is interpolate it freely and publish it under the guise of Madden ratings. I choose not to discuss all of the details on how I rate players because I want sole ownership over how I do things. I have put in thousands of hours into this project since 2009, and to be honest, I want the credit for that work. I am willing to discuss theory, but not specifics. Get what I am saying?

I think so.
You receive grades from a source and convert those grades to a Madden rating? I've got a background in statistical analysis so that part is fairly direct. And don't worry, I am not trying to move in on your action

I'm just wondering on things where that is not a obvious or direct statistical equivalent in what ways could you quantify it. Take route running. If you use Donny's system you just assign numbers at a whim. In what ways could you make route running empirical? Because depending on how the raw data collector operationalizes it the difference between your rating and Donny's rating may only be the distribution.

# 1210 ggsimmonds @ 07/22/14 12:47 AM

Quote:

Originally Posted by DCEBB2001


	Production never tells the whole story. Even great players can struggle to produce. Is Larry Fitzgerald a worse player now because he hasn't cracked 1000 receiving yards since 2011? That is why I don't trust stats. Stats don't tell me how fast, strong, agile a player is. They don't tell me how hard a QB throws a ball. That stuff is useless to me, and I don't/won't use it. I use the scouting data like every decent front office does.

Does not tell the whole story, but it does tell a good portion of it. What explains the drop in production for Larry?
"How fast, strong,agile...how hard a QB throws a ball"
Those are all physical though. Stats can tell you how skilled a player is in certain areas of the game. And I think it is fair to say that in the NFL the gap in physical ability is not quite as big as the gap in skills.
A WR's ability to gain separation, i.e. run routes, catch the ball, attack the ball at the highest point, get off the line, etc are greater predictors of success than 40times, cone drills, and the shuttle run.

The scouting data is derived from in game performance. And to me, in game performance falls under production. Production does not simply mean how many yards a QB threw for.

# 1211 DCEBB2001 @ 07/22/14 12:48 AM

Quote:

Originally Posted by ggsimmonds


	I think so. You receive grades from a source and convert those grades to a Madden rating? I've got a background in statistical analysis so that part is fairly direct. And don't worry, I am not trying to move in on your action I'm just wondering on things where that is not a obvious or direct statistical equivalent in what ways could you quantify it. Take route running. If you use Donny's system you just assign numbers at a whim. In what ways could you make route running empirical? Because depending on how the raw data collector operationalizes it the difference between your rating and Donny's rating may only be the distribution.

Statistical analysis is both my career and hobby. I do analysis at work 40-60 hours a week, then do it at home (to the chagrin of my GF) for a few more each day...but...I love it.

Basically, every attribute is on a 0.0-5.0 scale with detail down to the 0.1. The scouts create the grades, my source being one of them. The grades are then sent to me, privately for every player in their database (some 70,000 players as of today). So for a WR's ability to run a route, there is a grade for that that ranges from 0.0 to 5.0. Nobody in the NFL right now has that grade. The highest out there right now is 4.2. That means that there either no player has achieved that perfect 5.0 score, or a player who did achieve the perfect score is no longer on their radar in their present database (Jerry Rice maybe?...I'm not even sure). Now do that for each position for about 10-20 categories depending on the position. The categories are usually very specific. Some, I can't even use in Madden because Madden doesn't have an attribute for them.

There are several ways to interpolate the data, but right now the site uses points at the max, min, and average to determine the polynomial function of the curve. Those are then set to the proper range in Madden ratings (which is very difficult to determine and is often nebulous).

# 1212 charter04 @ 07/22/14 12:51 AM

Quote:

Originally Posted by ggsimmonds


	Does not tell the whole story, but it does tell a good portion of it. What explains the drop in production for Larry? "How fast, strong,agile...how hard a QB throws a ball" Those are all physical though. Stats can tell you how skilled a player is in certain areas of the game. And I think it is fair to say that in the NFL the gap in physical ability is not quite as big as the gap in skills. A WR's ability to gain separation, i.e. run routes, catch the ball, attack the ball at the highest point, get off the line, etc are greater predictors of success than 40times, cone drills, and the shuttle run. The scouting data is derived from in game performance. And to me, in game performance falls under production. Production does not simply mean how many yards a QB threw for.

Seems like not having Kirt Warner at QB and not having Boldin across from might make more of a difference than him not being as good

# 1213 DCEBB2001 @ 07/22/14 12:53 AM

Quote:

Originally Posted by ggsimmonds


	Does not tell the whole story, but it does tell a good portion of it. What explains the drop in production for Larry? "How fast, strong,agile...how hard a QB throws a ball" Those are all physical though. Stats can tell you how skilled a player is in certain areas of the game. And I think it is fair to say that in the NFL the gap in physical ability is not quite as big as the gap in skills. A WR's ability to gain separation, i.e. run routes, catch the ball, attack the ball at the highest point, get off the line, etc are greater predictors of success than 40times, cone drills, and the shuttle run. The scouting data is derived from in game performance. And to me, in game performance falls under production. Production does not simply mean how many yards a QB threw for.

What you are missing, however, is that it is not always correct that skill leads to production. We only use the raw physical measurables to determine the 5 raw physical attributes. Everything else is based on the data and its correlatives.

Performance and production are two very different things. A receiver can run a route and get wide open because he is a great route runner, but if the QB doesn't throw him the ball and throws to someone else, you won't know how good he is from the stat line. They don't tell the whole story. Moneyball doesn't work in football.

That said, I am growing tired of this discussion for tonight. I don't think you will convince me one way or the other, nor I you. So let us just leave it at that. If you use my ratings and rosters made from them cool. If not, well that's cool too. You can even make your own site based on stats if you like. Different strokes for different folks.

# 1214 DCEBB2001 @ 07/22/14 12:53 AM

Quote:

Originally Posted by charter04


	Seems like not having Kirt Warner at QB and not having Boldin across from might make more of a difference than him not being as good

Good, I'm not the only person who sees that.

# 1215 ggsimmonds @ 07/22/14 01:03 AM

Quote:

Originally Posted by DCEBB2001


	What you are missing, however, is that it is not always correct that skill leads to production. We only use the raw physical measurables to determine the 5 raw physical attributes. Everything else is based on the data and its correlatives. Performance and production are two very different things. A receiver can run a route and get wide open because he is a great route runner, but if the QB doesn't throw him the ball and throws to someone else, you won't know how good he is from the stat line. They don't tell the whole story. Moneyball doesn't work in football. That said, I am growing tired of this discussion for tonight. I don't think you will convince me one way or the other, nor I you. So let us just leave it at that. If you use my ratings and rosters made from them cool. If not, well that's cool too. You can even make your own site based on stats if you like. Different strokes for different folks.

Have a good night but I will leave you with this. We don't need to convince each other because I think we don't actually disagree. I think you correctly pointed out that performance is not the same as production. That is my error, I was kind of using them interchangeably. The reason I did that was because when a WR catches a pass that is not only his production. That is the result of a team.

For your rosters, if EA does the right thing and allows for customizable draft classes I probably will give them a spin.

# 1216 michiganfan8620 @ 07/22/14 01:07 AM

Quote:

Originally Posted by DCEBB2001


	Sorry you disagree, but I go with what the grades say. I don't make this stuff up. I am an expert in data analysis. I interpolate and publish.

I guess I just don't understand how a scout could grade a player so lowly based on the fact that he is suspended if you aren't factoring in production. How is he any less of a player? And another thing, I'm guessing Madden doesn't use historical players in their interpretation of the data. Thus, if a player was considered to have the strongest arm in football, they give him a high rating. Even though he might only be a 4.0 on the 0-5.0 scale, which I'm assuming equals a 80 rating, he gets a high 90 rating in order to fit the programming of the game (making sure he can throw the ball as far as he actually can in real life). And then they go from there with the rest of the players. They are just focused on the present time, and how players currently in the league compare to each other, not on how they compare to all-time greats. At least that's what I'm assuming Madden does.

# 1217 ggsimmonds @ 07/22/14 01:13 AM

Quote:

Originally Posted by DCEBB2001


	Statistical analysis is both my career and hobby. I do analysis at work 40-60 hours a week, then do it at home (to the chagrin of my GF) for a few more each day...but...I love it. Basically, every attribute is on a 0.0-5.0 scale with detail down to the 0.1. The scouts create the grades, my source being one of them. The grades are then sent to me, privately for every player in their database (some 70,000 players as of today). So for a WR's ability to run a route, there is a grade for that that ranges from 0.0 to 5.0. Nobody in the NFL right now has that grade. The highest out there right now is 4.2. That means that there either no player has achieved that perfect 5.0 score, or a player who did achieve the perfect score is no longer on their radar in their present database (Jerry Rice maybe?...I'm not even sure). Now do that for each position for about 10-20 categories depending on the position. The categories are usually very specific. Some, I can't even use in Madden because Madden doesn't have an attribute for them. There are several ways to interpolate the data, but right now the site uses points at the max, min, and average to determine the polynomial function of the curve. Those are then set to the proper range in Madden ratings (which is very difficult to determine and is often nebulous).

(Respond to this whenever you wish, not trying to hold you hostage)

I am entering graduate school at UMD College Park studying Public Policy. So I am still something of a duckling learning to fly when it comes to statistical analysis.

But for this, I think my interest is more of a scouting perspective. What goes into a grade in route running? Number of steps/yards covered, getting off the line, breakdown (how many steps do they take), foot planting, things like that?

what about coverage skills? It seems easier to grade someone running routes as you would have a slightly easier time doing that without a defender than measuring coverage skills without a WR.

# 1218 JHTJ @ 07/22/14 01:28 AM

I couldn't even make it past the Dolphins before I saw some serious issues with the ratings on that site. I mean, real bad. Now I’m not suggesting players as a whole are underrated, or the team is collectively better than depicted at all. I could care less, honestly. I will say though that rating Pro Bowl C Mike Pouncey as the 4th worst player on the team is a travesty if Ive ever seen one. Rated 53 behind two undrafted Centers fighting for roster spots is crazy. You have an undrafted ROOKIE rated ahead of All SEC, All American, Pro Bowl, PFF rated top 10 Pouncey.

I'm sure undertaking and implementing this rating process is a difficult task so this post comes with all respect due. I might be willing to concede that considering his injury pouncey had a slightly off year, but I watch every game - Every snap my Fins play repeatedly and I can say unequivocally that rating is garbage. I don't expect you to defend this rating because, quite honestly, i don't think it's possible. The fact is its your site and you can publish what you like. I just wanted to point out what I saw as a glaring inaccuracy.

# 1219 michiganfan8620 @ 07/22/14 01:35 AM

Quote:

Originally Posted by JHTJ


	I couldn't even make it past the Dolphins before I saw some serious issues with the ratings on that site. I mean, real bad. Now I�m not suggesting players as a whole are underrated, or the team is collectively better than depicted at all. I could care less, honestly. I will say though that rating Pro Bowl C Mike Pouncey as the 4th worst player on the team is a travesty if Ive ever seen one. Rated 53 behind two undrafted Centers fighting for roster spots is crazy. You have an undrafted ROOKIE rated ahead of All SEC, All American, Pro Bowl, PFF rated top 10 Pouncey. I'm sure undertaking and implementing this rating process is a difficult task so this post comes with all respect due. I might be willing to concede that considering his injury pouncey had a slightly off year, but I watch every game - Every snap my Fins play repeatedly and I can say unequivocally that rating is garbage. I don't expect you to defend this rating because, quite honestly, its impossible. The fact is its your site and you can publish what you like. I just wanted to point out what I saw as a glaring inaccuracy.

He will say its due to Pouncey's injury and surgery, and that once he is back to health, his rating will promptly be adjusted. Although I noticed something there as well. Why is Dion Jordan's rating a 72 when he is suspended, whereas Robert Mathis is a 44 with the same length suspension?

# 1220 JHTJ @ 07/22/14 01:48 AM

Quote:

Originally Posted by michiganfan8620


	He will say its due to Pouncey's injury and surgery, and that once he is back to health, his rating will promptly be adjusted. Although I noticed something there as well. Why is Dion Jordan's rating a 72 when he is suspended, whereas Robert Mathis is a 44 with the same length suspension?

My apologies, I would have liked to spend some time reading the thread but with 130+ pages I decided that just wasn't happening. The obvious question would be why are ratings being tweaked in the offseason based on injuries with speculative impact on performance? Eh, I'm sure that too was answered at some point in time. Disregard, if I really want the answer ill dig for it. I will say that regardless of justification, ratings adjustments made based on injuries or suspensions doesn't sound like a system that interest me.

« Previous 1 2…58 59 606162 63 64…76 77 Next »

More Links

FBGRatings Seeks to Recalibrate Madden Ratings, Change Game