FBGRatings Seeks to Recalibrate Madden Ratings, Change Game

[DCEBB2001]

Quote: Originally Posted by charter04

I'm more of a fan of equal interval. Either way can't really be used in CFM as is anyway. Until there is a draft class share on madden share. I know guys get all weird about overall ratings but, I just like the equal interval best.

Equal interval is by far the most ACCURATE method, especially for the interpolation of the scouting data. For instance, if the RET attribute scale is from 5.0 to 0.0 and the Madden range is 100 to 0, then you can determine that each 0.1 is equal to 2 points. Which makes sense, because if the scouting data says that Player A with a RET of 0.2 is twice as good at returning kicks and punts than Player B with a RET of 0.1, then that comes out to be a RET rating of 4 for Player A and 2 for Player B.

The problem is that people totally freak out when they see how low the OVR ratings are....starters in the 60s. The version currently on the website is more traditional-FBG, which is closer to current Madden, but utilizes old FBG metrics. I personally love the equal interval method, but once again, when I did it last year, people freaked out a bit too much.

[charter04]

Quote: Originally Posted by DCEBB2001

Equal interval is by far the most ACCURATE method, especially for the interpolation of the scouting data. For instance, if the RET attribute scale is from 5.0 to 0.0 and the Madden range is 100 to 0, then you can determine that each 0.1 is equal to 2 points. Which makes sense, because if the scouting data says that Player A with a RET of 0.2 is twice as good at returning kicks and punts than Player B with a RET of 0.1, then that comes out to be a RET rating of 4 for Player A and 2 for Player B. The problem is that people totally freak out when they see how low the OVR ratings are....starters in the 60s. The version currently on the website is more traditional-FBG, which is closer to current Madden, but utilizes old FBG metrics. I personally love the equal interval method, but once again, when I did it last year, people freaked out a bit too much.

I really don't want to do all the teams already done over again but, I like the 360 roster I have off madden 25 done in the equal interval. I'm cool with what ever. I'm just going to try and get this new roster done before a possible overhaul. Lol

[Hooe]

To provide an additional point of data, if I were to use these ratings I would want to use them in Connected Franchise mode (assuming editable draft classes gets added into the game). To that end, I think that reducing the OVR rating too low might screw with the game's contract logic, and I'd be reluctant to take these ratings into that mode and therefore use them at all (franchise mode is where I spend the overwhelming majority of my time with Madden).

However, given the current lack of editable draft classes and thus no way to ensure that the ratings philosophy is maintained throughout the life of a franchise, I don't have a preference between ratings methods.

[DCEBB2001]

Quote: Originally Posted by CM Hooe

To provide an additional point of data, if I were to use these ratings I would want to use them in Connected Franchise mode (assuming editable draft classes gets added into the game). To that end, I think that reducing the OVR rating too low might screw with the game's contract logic, and I'd be reluctant to take these ratings into that mode and therefore use them at all (franchise mode is where I spend the overwhelming majority of my time with Madden). However, given the current lack of editable draft classes and thus no way to ensure that the ratings philosophy is maintained throughout the life of a franchise, I don't have a preference between ratings methods.

All of this really does hinge on the editable draft classes. The point you raise about the contracts is valid as well. That was one of the reasons I went with a traditional FBG system this time around. Each method has its own pros and cons. One of the cons of the traditional FBG ratings is that some position-specific attributes are quite low, like the average CTH of a WR/TE being 67 instead of 78 or so as it was rated in M10.

TBH, M10 probably had the best ratings system since M2003, IMO. Using M10's system as a base would be more in line with how EA rates players now, but you would get the dreaded rating over-inflation where an attribute like ACC averages 79....well above the averages for the other physical attributes (SPD 73, JMP 66, AGI 74, STR 71).

In fact, according to the data, they should all be rated on the same scale either with the same average rating (like 70 or 75 or something like that) or utilize the equal interval (where the space between each point is the same). When you use the equal interval, however, your average STR in the league goes from 70 (if you set it) to 44! Believe it or not, players are far more fast/agile/quick than they are strong on the whole, if all interval points are equal. I imagine that having more WRs/DBs on rosters than say Cs plays a role in bringing the average down as well, but that is a pretty significant drop.

This dilemma is constant. How do you accurately rate players. Outside of the game, the equal interval (let the chips fall as they may) is the most accurate. Inside the game, the M10 and M25 methods are the most accurate for getting the OVRs to be comparative to their attribute parts. The best gameplay, from opinions expressed by several users, is with the traditional FBG or Equal Interval methods.

So as you can see it is a figurative "what'll it be?" for each gamer. I expressed on this forum that if 2K throws a hat into the NFL Football Gaming ring, I would love to throw my hat into THEIR ring and push for an equal interval approach, backed by real data, of course. That could, in the future, be the real wild card. Madden's way of rating players is broken, but we have no (other) choice but to use their flawed system to rate players.

Maybe we just have to rate players crappily (sp?) in their crappy system so that we can use franchise modes and draft classes effectively. For now, however, I think we should wait to see if EA will allow us to edit draft classes. If they do, it could be full-on for utilizing a better ratings system in-game.

[Fdiez]

Brilliant work, thank you.
It will be the first time i'll try your roster data so i can't make any comment about the rathing method.

[DCEBB2001]

Quote: Originally Posted by Fdiez

Brilliant work, thank you. It will be the first time i'll try your roster data so i can't make any comment about the rathing method.

Sounds good. Feel free to ask any questions via PM or this thread.

Unlike EA, I am more than happy to share some insight as to what I do on the site.

[Hooe]

Process question - how much do you weight your scouting data vs player performance on field over long stretches of time?

For example, Joe Flacco. All of his accuracy ratings approach 90. However, his true accuracy percentage on Pro Football Focus (which corrects for dropped passes, passes tipped at the line of scrimmage, spikes, and passes where the quarterback is hit as he's thrown), Flacco finished 25th amongst all qualifying QBs in 2013 and finished in the 20s in the 2012 and 2011 seasons as well. Among active QBs, he's 17th in career completion percentage.

How does your system weigh that against his high accuracy ratings which I presume are drawn straight from your scouting data? What am I missing from the equation here? Should I be looking instead at the Ravens' receivers not consistently running great routes and Flacco missing them on that accord and/or a poor pass blocking offensive line resulting in more throws under pressure?

Again, I ask not as a criticism but for an understanding of process with a specific example that jumped out at me.

[DCEBB2001]

Quote: Originally Posted by CM Hooe

Process question - how much do you weight your scouting data vs player performance on field over long stretches of time? For example, Joe Flacco. All of his accuracy ratings approach 90. However, his true accuracy percentage on Pro Football Focus (which corrects for dropped passes, passes tipped at the line of scrimmage, spikes, and passes where the quarterback is hit as he's thrown), Flacco finished 25th amongst all qualifying QBs in 2013 and finished in the 20s in the 2012 and 2011 seasons as well. Among active QBs, he's 17th in career completion percentage. How does your system weigh that against his high accuracy ratings which I presume are drawn straight from your scouting data? What am I missing from the equation here? Should I be looking instead at the Ravens' receivers not consistently running great routes and Flacco missing them on that accord and/or a poor pass blocking offensive line resulting in more throws under pressure? Again, I ask not as a criticism but for an understanding of process with a specific example that jumped out at me.

Every player is graded as an individual only. Their abilities are what are reflected by their scores. The cause and effect of having other teammates on the field are not taken into account. What that means is, in a perfect, static, environment, how accurate is Flacco (per your example) at throwing short (<10yds), medium (10-25yds), and deep (25+yds) passes.

Statistics play absolutely no part in any of this. It is all based on the player's individual traits based on what the scouting department that provides the data says. I mention this in the FAQ section of the website as well with another example. The reason for this is because any one of a million factors could affect Flacco's throw or the reception on any given play. That is not a fair measure of traits. They need to be calculated statically. Then, once every ability is graded and accounted for, you throw them all together.

This approach throws production and statistics out of the equation. To quote "that head coach" in New England:

"We are NOT in the business of scouting for PRODUCTION; we are in the business of scouting usable TRAITS".

An example. Joe Flacco. By himself, we know he has a pretty big arm and has some decent accuracy along with some mobility issues. Standing alone in the pocket he can deliver some very nice passes. Now free-blitz 3 LBs at him. Good luck seeing comparable accuracy on those throws. The cause in having three LBs chasing after him caused his accuracy to dip as he was forced to throw on the run. That doesn't mean he is less accurate does it? Of course not. That would be a logical fallacy. What it does mean is that his lack of mobility and ability to throw while evading pressure DOES affect his accuracy.

It is all about causation versus correlation. Sites like PFF don't take into account CAUSATION, but rather, look at mere CORRELATION. There is a big difference as correlation doesn't necessarily suggest causation.


Make sense?