Madden NFL 15 - Player Ratings Visualized

[richrolla]

This is awesome CM! Is there anyway for you to post the numeric means that fit each individual color code?

I'd love to use that in unison with the weekly roster updates to see when a player improves or regresses into the next category.

[CM Hooe]

Quote: Originally Posted by richrolla

This is awesome CM! Is there anyway for you to post the numeric means that fit each individual color code? I'd love to use that in unison with the weekly roster updates to see when a player improves or regresses into the next category.

It'll take me a bit to do (the way I saved out those sheets, the formulas weren't preserved), but I'll see if I can't have that up sometime late tomorrow.

[richrolla]

Quote: Originally Posted by CM Hooe

It'll take me a bit to do (the way I saved out those sheets, the formulas weren't preserved), but I'll see if I can't have that up sometime late tomorrow.

That would be awesome dude. I could try doing some myself, but I definitely slept through most of my advanced psych stats courses so there's no telling if I get it right.

[CT Pitbull]

Quote: Originally Posted by CM Hooe

Not sure if this warrants its own thread but I didn't see a thread to discuss all the player ratings in totality so here goes. If you're like me, you learn better and process information more quickly when things are presented visually. A spreadsheet full of numbers is an incredibly dense thing which is hard to gloss over. Thus, quickly making comparisons between players in the Madden NFL 15 ratings reveal was a bit of a painful exercise. I made the following document in an attempt to aid myself to this end. First thing I did was to group together all the players at each position. Then, for each rating for each player (divided into individual position groups), I calculated the cumulative normal distribution and finally color-coded the result. Color key is as follows: Name Color Definition Description Elite r ≥ μ+2σ Player's rating is at least two standard deviations above the mean rating for this category and position; i.e. the player has a great rating in this category relative to peers at his position. Good μ+σ ≤ r < μ+2σ Player's rating is at least one standard deviation above the mean rating for this category and position; i.e. the player has a good rating in this category relative to peers at his position. Average μ ≤ r < μ+σ Player's rating is at or above the mean rating for this category and position; i.e. the player has an average rating in this category relative to peers at his position. Subpar μ-σ ≤ r < μ Player's rating is below the mean rating for this category and position; i.e. the player has a below-average rating in this category relative to peers at his position. Poor r < μ-σ Player's rating is at least one standard deviation below the mean for this category and position,i.e. the player has a poor rating in this category relative to peers at his position. Included in the ZIP are two PDFs - one sorted by OVR per-position, and one sorted by team then OVR per-position. Hopefully at least one person finds the product of my boredom useful, haha.

This is the kind of effort EA needs to have on their payroll. Some one with a creative and "new" way to look at ratings. Hell of a job man.

[Sheba2011]

Quote: Originally Posted by Big FN Deal

Not trying to be funny but that's the point of it, if you can use scouts or whatever to decipher which player is better/higher rated what's the point in hiding ratings at all?

An in game scouting report like anything else would be subjective. It's not going to say Player A is better than Player B and here is his ratings. It would simply list his strength's and weakness and you as the GM would need to decipher based on all the information provided which player will be better. Like in the real NFL sometimes GM's are right and they select Marcus Allen, other times they are wrong and they select Darrin Nelson (selected 3 picks before Allen).

[Big FN Deal]

Quote: Originally Posted by Sheba2011

An in game scouting report like anything else would be subjective. It's not going to say Player A is better than Player B and here is his ratings. It would simply list his strength's and weakness and you as the GM would need to decipher based on all the information provided which player will be better. Like in the real NFL sometimes GM's are right and they select Marcus Allen, other times they are wrong and they select Darrin Nelson (selected 3 picks before Allen).

Spoiler

Again, when I state "scouting" I'm referring to having actual in-game scouts or like it is now in Madden where we have to "unlock" certain details, I think that's unnecessary. I recently got a chance to watch the movie Draft Day and it showed a good depiction, imo, of all teams having the same info, they just had to interpret it in their own way. Of course a video game doesn't have game film, so I'm all for teams applicable staff, including the GM and Owner, having a separate OVR rating/color code, to represent their evaluation of each player.

NFL teams aren't "scouting" players that have played in the NFL, they scout unproven players, like draft classes, walk ons, practice squad players on other teams, etc. Like I stated earlier,, in the case of those type players, there needs to be a wider range to the ratings/color code to represent larger uncertainty but again the individual OVR rating/color code for each applicable staff would be their personal evaluation of those players, not some way to pinpoint or narrow down what a players true OVR rating is under the hood.

I put the first part in a "spoiler" because I feel like I just keep repeating the same thing but not getting my point across about scouting. To try to simplify, I disagree with the color code having anything to do with the accuracy of individual ratings because what it would lead to is people just trying to put together the most accurate "scouts" or staff to try to decipher actual ratings, imo. I think it should be like it is now, just a universal system that displays ratings ranges versus exact numerics. I feel the evaluations or scouting some want can be accomplished through a color coded OVR rating for each player, from each applicable team staff. That OVR would then be just each staff members personal perception/evaluation of those players like in real life, not some formula to narrow down what a players true OVR is under the hood.

[CM Hooe]

Upon request, I've posted statistics for all ratings at all positions. Find a new PDF document in the original post.

Definitions refresher:
MEAN - the average value of all the data in the set.
MEDIAN - the "middle" value of all the data in the set after the data is sorted. Within the set, there are just as many values less than or equal to the median as there are greater than or equal to the median. Not an average.
STDEV - standard deviation; amount of variation of all the data in the set. A higher value means that the values are spread further apart, a low value means the values are all tightly packed together.

So for example, the average AWR rating for all the quarterbacks in the game is 65. If your QB has AWR > 65, he's has above average AWR (color-keyed yellow in my original document); if not, he's got below average AWR (color-keyed orange in my original document). Standard deviation for AWR is 15, so the vast majority of all the QBs will have a AWR value greater than or equal to 50 and less than 80. If a QB has at least 80 AWR (one full standard deviation greater than the mean), he's markedly better than his peers (color-keyed green). If a QB has at least 95 AWR (two full standard deviations away from the mean), he's an extreme case and has AWR which is likely the best in the league (color-keyed blue), any replacement QB you may field instead will most likely have worse AWR. (For completeness: AWR below 50 gets color-keyed red.)

[WickedFastball]

Quote: Originally Posted by Sheba2011

An in game scouting report like anything else would be subjective. It's not going to say Player A is better than Player B and here is his ratings. It would simply list his strength's and weakness and you as the GM would need to decipher based on all the information provided which player will be better. Like in the real NFL sometimes GM's are right and they select Marcus Allen, other times they are wrong and they select Darrin Nelson (selected 3 picks before Allen).

I think what we really need in CFM is a REAL 1 or 2 paragraph scouting report using terms such as "Quick feet" and "Soft Hands" or "SYSTEM QUARTERBACK"!!! (Which really means-NOT compatible for the NFL) side note Tom Brady was called a "System QB" coming out of Michigan, which brings me to my next point.
The more money you spend on your scouts OR upgrade your teams scouting abilities?!?!, the more "accurate" your reports should be.
Saving money on scouts could = thinking Marcus Allen was your guy. OR Tom Brady ISN'T just a System Quarterback!