Monday, September 10, 2012

Madden 13 player ratings - Has EA and Donny Moore created a monster? Weekly "kneejerk" player rating adjustments have to stop.

You may be thinking, "what monster?"  The monster I am referring to is YOU, the Madden NFL fan who expects player rating adjustments after every good or bad game.  Even worse, you might expect to see an attribute adjustment after just one good or bad play.  Why do so many Madden fans expect this type player rating system?  There is a simple answer.  Over the years, Donny Moore and EASports have conditioned fans to expect this way of rating players.

In week 1 alone, Donny Moore has tweeted this:

Dennis Pitta having a nice evening  +++

Two things are certain this week:  & Ogletree getting some extra ratings love; Brandon Weeden getting some extra scrutiny

Wow Jay Cutler, wow. Awareness - - -

Sorry David Wilson,  Carry rating - - -

Now you might say, "those are not bad ideas". My question is why not wait three to four weeks before making attribute adjustments.  In my opinion, a more patient approach would result in more accuracy and consistency for ALL player ratings.

Just last night Peyton Manning stated, "It's just one game, I try to keep it in perspective."  That was after beating Pittsburgh (31 - 19) in his first regular season start in over a year.  This is the attitude I would like to see taken in regards to Madden player ratings.

Making ratings adjustments after one game (or play) is one of the biggest problems in player ratings.  These adjustments are based on a meniscule sample size, which results in inaccurate attribute rating that may or may not be corrected later in the season.

Some people will say I am "hating" on Donny Moore and EA.  I am actually trying to help them.  They are not the only problem.  The expectation of Madden fans is also contributing to the problem.

Here are some tweets from Madden fans to Donny Moore:

 Hopefully Gabbert/Monroe/Shorts get some ++++?

 Entire Bengals defense: play recognition-------

 Alfred Morris  +++

 also, Cary Williams and Justin Tucker should get some sort of increase as well. 

 joe flacco deep pass accuracy +++

Do those type of tweets look familiar?  I could list pages of this kind of tweet to Donny.  The fans are even speaking in the same language as Donny Moore.  To be far, I use to do this exact thing in previous years.  I believed, those type of tweets were the only way to help my favorite players get more accurate player ratings.  Then, last November, I realized that I was contributing to the problem.  Not only did I become increasing frustrated with inaccurate attribute ratings for my favorite team, but I found myself digging into all attribute ratings for all teams.  This was the beginning of my blog and my goal of pushing EA and Donny Moore to produce more accurate player attributes for all players.  When I say "all players", I mean that.  Not just the popular players, or the players who played in primtime, but every player good or bad.

Believe me, if I can change, you can change.

Let me be clear, fans are just as much at fault as Donny Moore when it comes to the inconsistency in Madden player ratings.  You may not believe that, but take a step back and think about it.  Deep down, you know it's true.

Consider this, Donny's tweet about rookie RB David Wilson, suggests he will decrease his carry (CAR) attribute.  He fumbled early in the Wednesday night game and was a none factor the rest of the way (his coach was teaching him a valuable lesson by benching him).   Wilson had issues in college, he had a fumble per touch rate of 1.6%.  Donny gave him a very accurate initial carry attribute of 79.  The Madden 13 average CAR attribute for a RB is 80.  So, Wilson is already below the Madden average.

It's a fact that his career did not get off to a great start, but would it be horrible to let him stay at the 79 CAR attribute and see how things play out the next 3 weeks?  Is Donny going to be tracking his every carry from now until the end of the season?  What if Wilson goes 100 touches without a fumble, is Donny going to have the time to notice and readjust his CAR attribute.  I don't think so; Donny is a very busy man.  Now if you waited to decrease his CAR attribute and tracked what happens the next 3 weeks, it's possible no decrease is necessary.  Therefore, Donny doesn't have to go back and fix it.  This saves time, and time is something Donny doesn't have much of from week to week.

I give Donny a ton of credit, he did wonders with both the RB an QB carry (CAR) attributes.  It took me several hours to accumulate the statistics and provide suggestions to EA for those CAR attributes.  Those suggestions were not based on one play or one game, they were based on years of performance.  I applaud Donny for using that data and applying it (here is a blog showing that  Click).  After making such great progress toward attribute accuracy, why go back to jumping the gun?

Get this, going into the Monday Night Football games, there were 46 fumbles (not all were lost) in the NFL during week one.  Is Donny Moore going to treat every one of those fumbles equally?  I don't think so.  I plan on updating my fumble percentage statistics at both mid point and the end of the NFL season, hopefully that will help EA.  I don't believe a carry (CAR) attribute should be decreased after one play or one game.  I believe the larger the sample the size the more accurate the ratings will be.

I just wanted to take a moment and expression my thoughts with this blog.  I know many people will disagree with me and that's ok.  EA and Donny have created a monster, and before real player rating accuracy can be attained, the monster must be slayed.  The change has to start now.  It's going to be a bumpy road, but I believe it can be done.  Does this mean only 4 or 5 roster updates will take place during the season.  NO, it doesn't; not every player will be on the same schedule (due to injury, playing time, suspension, etc.).  Weekly roster updates will take place, but the ups and downs of individual player ratings will be reduced and result in more consistency and accuracy.

Currently, one bad play or game can result in attribute decreases (the opposite for increases).  Unfortunately, many players will fall through the cracks after a decrease (or increase) and will be forgotten as the season progresses.  What if that was the worst game or play he had all year?  How much should the player pay for that poor performance and at what cost?

If you have followed me for any period of time, this is not a new suggestion.  I will continue to beat this idea into the ground because I truly believe it will result in a better Madden NFL game.


Michael Mann (former "MONSTER")


  1. First I'd like to say that I agree that a lot of things need to be based on performance over time instead of 1 play or game. On the other hand, we are playing M13 now not M11-13, and I'd prefer to be playing a game that represents how players are performing this year. By the time we get to week 14 those vets who were living off last years ratings need to be dropped if they're not performing. Breakout players need to be recognized and rewarded.

    1. I agree Vets shouldn't only live off of the past, but the past can't be forgotten either. A 4 week window for evaluating ALL players will address Vets and Rookies a like. "Kneejerk" and Highlight based player rating does not equal "how players are performing this year". It's how they perform in the highlight or in primetime. Based on your comment, players would be completely different week to week with no regard for past performance. You would punish or reward them based on an extremely small sample size. I appreciate your point of view, but completely disagree. I think 4 week evaluation periods throughout the season will work for ALL players (rookies, breakouts, and vets).

      Question - Do you think one good game should count toward attribute ratings 4 weeks later? What if the other 3 games were average to below average for that player? No change would have been necessary in that case.

      Again, I know not everyone will agree with my ideas, but I appreciate you following the blog. Thanks.

    2. Was meaning more that I'm fine with taking 4 weeks to evaluate a player and see how it compares to their 3 yr average. But if they're consistently outplaying other players at their position over a couple 4 week sections then they need to be rewarded by being rated higher than those other guys. Same thing with those who underperform compared to what they're "supposed to do" based on their rating.

      As for your question, yes I would like it if 1 bad and 3 good were treated the same as 3 good and 1 bad, as an average.
      Think I told you in a twitter conversation once, it's the non ratio ratings that are the ones that can be changed based off one game imo. One example from last year is CB Chris Houston ran 4.32 combine if I rmbr right, RB A Peterson was closer to 4.4, AP was 98spd last year Houston 93, but in one game Houston chased him down from behind going up the sideline. That's an example of speed ratings being completely wrong in game and a real life example of proof the other way. That is something that should be addressed based off one play, because it's a fix of what is already wrong.
      Something else could be SPC. It's a subjective rating anyway so if in one game someone goes and makes 2 crazy catches then that could be something that's bumped up. Sure it would be nice if there was someone who kept track of attempted SPC stuff, but I don't know of anyone who does. There's no way (yet) for someone to prove that a player is good at making SPC, and not just someone who attempts 10 and catches 2 and gets all the credit for the 2 and none for the others.

      One last thing, and this is what I believe is the hardest thing for whoever is doing the ratings, is team situation stuff. There are some really great players out there who are stuck on terrible teams who sometimes misuse their talent. Then when they don't produce statistically they get their ratings dropped even though it wasn't like the player really got worse. Take Reggie Wayne last year, he showed yesterday that he's still a great WR, yet his ratings dropped a ton last year because he played for an inept team. All of this is why, like many have said already, OVR ratings need to be detached from other ratings, same with team ovr ratings.

  2. From what I've seen and surmised from last years madden rating debate, the problem isn't with the frequency of the updates but rather with Donny Moore being restricted by the engine that calculates the overall rating. For example, Frank gore had a low pass block rating last year despite being regarded as one of the best pass blocking HBs in the nfl. Why? Because raising his blocking stat would raise his overall to a point where his overall was higher than hbs that were clearly better than him. So, his blocking stayed low. This wasn't missed it was just a decision by the madden team to leave a rating inaccurate rather than skew the overall or lower another attribute to balance out the overall. Changing the update frequency would have no effect.

    Also, I don't think Donny's changes qualify as knee jerk because they are rarely more than a point or two per week, hardly a drastic change. I would be more worried about him making large changes once every 3-4 weeks and being stuck with them for that long rather than living with a somewhat inaccurate rating for just a week or two.

    finally, you point to the fact that donny can't keep track of everything, but isn't that the point of the debate? Its basically using crowd sourcing to identify errors and correct them. Donny isn't going to increase anything just because people on the debate told him to, rather he looks at what people are talking about and investigates the top issues further. He has said as much on podcasts. Participating in the debate is actually helping, not hurting and minor weekly updates give more chance for feedback and corrections.

    1. Actually, PBK, PBF, PBS have no impact on a Running Backs OVR. See here:

      So, that is not preventing Gore from having better blocking attributes.

      EA did not choose to have Gore's blocking attributes inaccurate, they just don't have the resources to follow everything and although Niner fans have been screaming about it since last year, it still hasn't been corrected.

      The whole point of a 3 to 4 week evaluation is to increase attribute accuracy, therefore you wouldn't have to live with inaccuracy. I think you might be missing the point.

      As far as the increases go, you may only see a point or two change in OVR, but you could see a significant boost/decrease in a individual attribute. I suggest you look deeper into attributes and investigate the formula a little more, you will see many attributes have little to no impact on OVR.

      Thanks for the feedback, but I have to disagree with most of it.

    2. I guess the fundamental point I don't understand and haven't seen explained is how increasing the time between ratings increases the accuracy. Its the same amount of data being analyzed, 1 weeks worth over a week or 3 weeks worth over three weeks. Its the same people doing the evaluations. So, what exactly would change? Why would it suddenly be more accurate?

      If 49ers fans have been screaming for a whole year to get a rating changed and it hasn't, what makes you think they would change it if updates were less frequent? Whatever the reason, it seems EA has decided not to increase the rating, but I seriously doubt that Gore's rating is low just because Donny hasn't gotten a chance to look at it again. Of course, I'm using Gore as an example, but how do we know the reasoning behind any of the perplexing attributes? Just because EA hasn't explained it doesn't mean there's not a reason and it doesn't mean they've simply missed it.

      Until we know more about why they do what they do, how can we possibly hope to make accurate suggestions to improve their process? What I do know is that those that do know the process have asked for our opinions on which ratings are inaccurate via the weekly Madden ratings debate and that seems to be the best way to contribute for now

    3. It's about having more time and making less unnecessary changes. Like I said, part of the problem with the "kneejerk" ATTRIBUTE changes is that it's typically not the norm (in regards to the player's performance) and the player is giving an unrealistic attribute. Many times, this attribute is never corrected later in the year. Utilizing a watch list and slowing things down to every 4 week will help the player ratings person manage their time, IMO. The four week evaluation would also help put those highlight plays (good or bad) into perspective (in regards to the players overall ability.)

      Is it possible that ratings and attributes will be just as bad with a 4 week evaluation period? Yes, if the same person is doing the ratings they might not get any better. But, does it hurt to give it a try? It's all changing the way people think about roster updates. Granted, in the end, it's up to Donny Moore and EA to make that change. Getting rid of weekly "kneejerk" attribute changes is the first step in my opinion.

      For the fan that just wants their favorite team to get increases and doesn't care about attribute accuracy, this idea wouldn't be something they would want.

      I understand you disagree and respect that. I'm sorry you don't see how this could help. Thanks for following the blog BrookTrout.