What is "average"?

[KingV2k3]

Quote: Originally Posted by DCEBB2001

A lot of folks who write in about the site are not used to lower physical abilities and higher AWR ratings.

The current EA rating system inflates things that can easily be determined by "measurables" and devalues AWR...

That's EXACTLY why there's so many gameplay issues triggered by SPD, STR, JMP, ETC. and also why players rated below about 60 AWR are useless...

The current system creates super human lug heads...

Purplepower_NC did / is doing a NCAA12 roster based off some conclusions that Gotmadskillzson came up with that do exactly that...

Here's a link:

http://www.operationsports.com/forum...on-ncaa12.html

In that system, the lowest AWR is 70...which has a huge positive impact...

Make the players smarter and less "jacked"...

Plays great / works wonders...

I have no doubt that same principles would work for Madden...

[DCEBB2001]

Quote: Originally Posted by KingV2k3

The current EA rating system inflates things that can easily be determined by "measurables" and devalues AWR... That's EXACTLY why there's so many gameplay issues triggered by SPD, STR, JMP, ETC. and also why players rated below about 60 AWR are useless... The current system creates super human lug heads... Purplepower_NC did / is doing a NCAA12 roster based off some conclusions that Gotmadskillzson came up with that do exactly that... Here's a link: http://www.operationsports.com/forum...on-ncaa12.html In that system, the lowest AWR is 70...which has a huge positive impact... Make the players smarter and less "jacked"... Plays great / works wonders... I have no doubt that same principles would work for Madden...

As it stands currently, I have about 2% of the total database (400 players) with an AWR over 90. AWR is also scouted and is on a scale that currently deviates from 4.5 to .1. Note that the scale initially goes from 5.0 to 0, but the highest player in the database has retired and tragically died this summer (Seau). His AWR grade was a 4.8.

What do you all think about utilizing a hybrid approach to the ratings?

In this hybrid, the raw attributes (STR, SPD, ACC, AGI, JMP) would have upper and lower bounds set to 99 and 1 with all of the averages set to 70. For STR, I have some great data that reflects the 1RM for a player's squat and bench. You would be surprised how much some of these guys can squat for their positions (yes, some kickers can out-squat some OL, believe it or not).

The hybrid portion would come in where all of the dynamic skills would reflect something similar to how EA evaluates dynamic skills. This would allow for some overlap with the system provided for in the game, while still maintaining the integrity of a true "average" being considered in the raw attributes.

Or am I to understand that setting an upper/lower bound and average for the skills as they pertain to a postion is still the way to go.

I raise this question because it is hard to determine what the average for a dynamic skill should be.

An example: BCV for RBs.

EA uses a range of 98-42 with an average point of 79 for BCV in Madden 12.

FBG dictates that this range should run from 99 to 41 with an average of 70, as currently represented by the ratings presently on the site.

FBG has increased the upper bound, lowered the bottom bound, and lowered the "average" for a ball carrier to see the field and blocks developing.

The range for this attribute from the scouting data runs from 4.5 (MJD) to .1 (several players). The average for this data is 2.2. To me, I like the present approach that avoids the over-inflation of the BCV rating. An average RB with an average score of 2.2 being at a 70 BCV allows the guys at the top to really stand out. While the average for EA's ratings for the same attribute, 79, makes the attribute more top heavy. In essence, you would have more players at 70+ with EA's way of determining an average and fewer players at 70+ with the FBG system.

For the sake of variation and player differentiation from the mean, I prefer the FBG standard. What do you guys think?

[at23steelers]

@DCEBB: What do you think of what I said about #'s correlating to animations. Does 80 CTH means he catches the ball 80% of the time?? Is 90 THP means he throws the ball 70 yards? Isn't it pointless to create a player like Manning with a random # like 79 THP and then find out that means he can only throw the ball 30 yards?

[DCEBB2001]

Quote: Originally Posted by at23steelers

@DCEBB: What do you think of what I said about #'s correlating to animations. Does 80 CTH means he catches the ball 80% of the time?? Is 90 THP means he throws the ball 70 yards? Isn't it pointless to create a player like Manning with a random # like 79 THP and then find out that means he can only throw the ball 30 yards?

I agree, which is why incorporating an average that works well in the game could be vital. Hence the hybrid approach to utilizing EA's way of rating players for technical skills and correlating dynamic ratings.

[at23steelers]

Quote: Originally Posted by DCEBB2001

I agree, which is why incorporating an average that works well in the game could be vital. Hence the hybrid approach to utilizing EA's way of rating players for technical skills and correlating dynamic ratings.

Is there a way to time it though. I mean 70 sounds the best to me for speed. But let's say you're giving him 70 speed for his 4.8 40, but in Madden 70 speed means only being able to run a 5.0 in the open field? Is there a good way to test that. If we were just doing a game like NFL coach, then all this would be great, but it has to also work in the game of Madden. Hope that makes sense.

[SA1NT401]

Average should be 58-68 OVR.

In madden 12 I literally nerfed the entire league where players rated a 78 were now 68 etc.

Game plays so much better and player that should stand out do and players who are average played like it.

Just about every player aside from the top elite 25 should all have their ratings dropped by 10-12. What a huge gameplay differance it makes.

But can't edit anything in 13 so....

[DCEBB2001]

Quote: Originally Posted by at23steelers

Is there a way to time it though. I mean 70 sounds the best to me for speed. But let's say you're giving him 70 speed for his 4.8 40, but in Madden 70 speed means only being able to run a 5.0 in the open field? Is there a good way to test that. If we were just doing a game like NFL coach, then all this would be great, but it has to also work in the game of Madden. Hope that makes sense.

There IS a way to test it. We know the frame rate of the game (60 FPS). All you need to do is create a player with 70 SPD and say 70 ACC. Go into practice mode with the OFFENSE ONLY. Get the ball to the player with the 70 SPD/ACC. Line him up on the 40 yard line and run toward the end zone at full speed while keeping the player running straight. It works well if you try to run in the hash marks. Go into replay and adjust the angle of the camera to facing the side profile of the player from the run you just did. Export the video and see if you can get it onto your computer. Use a video editor to count the frames of the run from the time you start running to when you cross the finish line. Take the number of frames and divide that by 60 OR simply time the run in the editor by deleting excess frames before and after the run. Advanced editors will give you this ability. That should tell you how fast a player goes the first 10 yards and 40 yards in total. Then you can use the speed threshold to see if you can get 70 to be close to a 4.81 40.


I highly doubt that you will be able to get it exact. From what I have seen, the SLOW speed allows for the most realistic speeds on the field in relation to real life 40 times. However, I could be mistaken.

This really doesn't matter to me that much, however, because I am simply trying to get the RATINGS and their numerical values to better represent the verifiable data, not try to get it better in game. I don't bother with the in-game stuff because we already know that the player models vs. field size is off and players accelerate to top speed way too quickly as is. In essence, the game itself already limits the realism. I did do some work back in the day, however, with getting accurate runnig results for Madden 08 on PC by using the same process. Basically, if you wanted your fastest players to accelerate at the fastest recorded time for 10 yards (1.40s), you had to use 50 as the maximum for ACC. Then record each SPD+ACC combination (ie 45 ACC + 90 SPD = x 40 yard time) for every possible combination thereof. It was pretty sweet and the game played amazingly, but it screwed up the OVR ratings of course as most had ACC ratings in the 20s and 30s.

[KingV2k3]

Quote: Originally Posted by DCEBB2001

Or am I to understand that setting an upper/lower bound and average for the skills as they pertain to a postion is still the way to go

The ratings have to somewhat mirror EA's rating range, or I don't think they'd work in game.

Their "High End" is 99 and their "Low End" is around 40...

For the sake of easy "math", I'd call that a "Total" of 140, which when divided by the two elements makes "Average" 70...

I think you have to research the ranges for each attribute in each position and then "Bell Curve" the results..

I also think you have to make the ratings position specific or (for example) the Bills' Brian Moorman will out run and out tackle the other 10 guys every time out of the gate on special teams...



I doubt the game can handle a 1-100 scale, though I have to admit, I'm not positive...

As far as your BCV example goes:

That's the type of rating that will be enhanced or diminished when combined with that players AWR...

So, if a group of guys end up with the same BCV, their individual AWRs will separate them within that pack...

I really think you have a great system in development...

I'd stick to your FBG guns / instincts, rate a couple teams and then take 'em for repeated in game tests...

Be aware that this title has always played "truer" in Franchise Mode...tests done in exhibition games, play now games and practice mode are slightly off...

Anybody who grinds on slider settings will tend to agree...

Oh, and I agree:

"Slow" seems to be the most realistic to me and many others who 40 timed the heck out of this title...