After watching madden 17, i still feel like its ratings need to be spread

[Yazan Gable]

Which is why I said that a part of making ratings more palatable and understandable would be if the developers explained what data and film work they use to come to their conclusions. Use data from groups like FO or PFF to decide how the ratings shake out. Transparency and an understanding of the methodology would go a long way is what I'm saying.

[ConnSKINS26]

Quote: Originally Posted by Yazan Gable

Which is why I said that a part of making ratings more palatable and understandable would be if the developers explained what data and film work they use to come to their conclusions. Use data from groups like FO or PFF to decide how the ratings shake out. Transparency and an understanding of the methodology would go a long way is what I'm saying.

I like the spirit of your argument, but your examples wouldn't work because there are just as many problems with the underlying methodology of FO and PFF as you see in the current Madden ratings.

[Yazan Gable]

I don't offer them as solutions, but options for where to get the data. If we know where they are getting their data from (where we can also look them over) and then follow through their methodology for translating that data into ratings, I believe it would improve the discussion on how to distribute ratings in the roster. Even with the flaws that FO or PFF or any other group would have, at least we know what the data is and where it is from. If communication and methodology is clear between the developers and the players on what data is used and how it is used, you can have a much more fruitful discussion on how to best represent players through ratings.

[chunt04]

Quote: Originally Posted by Yazan Gable

I don't offer them as solutions, but options for where to get the data. If we know where they are getting their data from (where we can also look them over) and then follow through their methodology for translating that data into ratings, I believe it would improve the discussion on how to distribute ratings in the roster. Even with the flaws that FO or PFF or any other group would have, at least we know what the data is and where it is from. If communication and methodology is clear between the developers and the players on what data is used and how it is used, you can have a much more fruitful discussion on how to best represent players through ratings.

Yazan, this is exactly what I was bringing up in my "possible idea for community rosters" thread in the rosters section. If we could all agree on a methodology for data and statistics on these players, it would be possible for us to implement a roster for the community with a diverse range of ratings for players that is backed by statistical data. This would remove that "X player is better than Y player because I've seen them do this" type of argument and offer a great franchise roster for everyone. Everyone seems to be against the sites you've mentioned, like PFF, but I at least feel confident in the fact that they've amassed a bunch of data on players that we could use to at least have the ranking of players implemented with statistical backing. The ratings and scaling would have to be determined by us to make sure they work within the parameters for "tiered animations" and everything already in Madden. I would very much like to see the community start to work together on something like this.

What are some sites that people hold in high regard when it comes to statistical data for the NFL? I know Dan's site holds actual scouting information, does everyone feel that rosters should only be based off that data? What do you guys think?

[Yazan Gable]

I personally regard Football Outsiders highly (I was only using PFF as a familiar example) as they try to rely less on eye test and subjectivity and use statistics to back their assertions and analysis; this is not to say that scouting or film data has no place (my links to FO were to a writer who was doing film analysis of various QBs), but I trust them and view them more highly.

Taking scouting data from some source and not being able to really look over that data and how they came to that conclusion is obfuscating a sizable part of what we're trying to figure out in ratings. While things like the DVOA formula maybe proprietary, at least the numbers are still accessible to everyone.

[charter04]

Quote: Originally Posted by Yazan Gable

I don't offer them as solutions, but options for where to get the data. If we know where they are getting their data from (where we can also look them over) and then follow through their methodology for translating that data into ratings, I believe it would improve the discussion on how to distribute ratings in the roster. Even with the flaws that FO or PFF or any other group would have, at least we know what the data is and where it is from. If communication and methodology is clear between the developers and the players on what data is used and how it is used, you can have a much more fruitful discussion on how to best represent players through ratings.

The guy who does the ratings already said he uses PFF among other sources.

[CM Hooe]

Quote: Originally Posted by chunt04

What are some sites that people hold in high regard when it comes to statistical data for the NFL? I know Dan's site holds actual scouting information, does everyone feel that rosters should only be based off that data? What do you guys think?

It's my opinion that the data that Madden should be using to produce player ratings is stuff that isn't publicly available - not FO, not PFF, certainly not fantasy stats - but stuff that EA should have better access to by virtue of the NFL exclusive license.

Specifically I'm referring to professional football player scouting reports from sources such as BLESTO or National. These are the actual scouting organizations that the vast majority of NFL teams use as their baseline knowledge set for grading players (they supplement it with their own in-house scouting efforts). The common thread is that these sources evaluate a player individually and quantitatively with as little influence from game day performances as possible. Example, a receiver's "route running" grade wouldn't be determined at all from how often he was targeted (i.e. got open) in games, and it certainly wouldn't be determined by receiving yards or touchdowns. Rather the player is isolated as much as possible from the statistics - scouts study is release off the line, his footwork in his route stem, ability to use his hands to beat press coverage, ability to not tip his hand about the break of his route before he makes his cut... all the stuff happens before the receiver even has a chance to catch the ball, that's what determines a player's "route running" grade. Not stats, not analytics, just film study and assessment by watchful eyes trained to make such assessments.

The data set which powered Dan's FBGratings rosters, to the best of my understanding, doesn't come from either BLESTO or National, but comes from a similar professional football scouting source which grades players individually.

[chunt04]

I like that. Physical attributes and intangibles should be separate from the statistics achieved on the field. That being said, with Madden, it's necessary for us to make sure that the sim stats would play out closer to what we see as well. As long as the ratings we chose could provide good analysis and separation of players (like in CM's previous post), and provide realistic sim stats for CPU teams in Franchise, we would have ourselves a winner. I would only want to take on the task of creating a roster if it was backed by real life data, as I am not going to waste my time creating something that is purely subjective and a matter of opinion.

How did the sim stats play out for 1 season with last year's FBG roster?

Would you guys rather see a roster that focused on optimizing gameplay (through animations) and diversifying players in the league? Or would you rather see something holistic that also tried to encompass sim stats and tries to give the best of both worlds (obviously with some compromise on both ends)?