Tuesday, July 31, 2012

NEW Madden 13 player ratings are starting to be released by EA. My thoughts on the top 10 QB's.

8/3/12 UPDATE - EA released the full QB attributes for ALL QB's.  Here is the link:  http://www.easports.com/madden-nfl/news/article/complete-qb-ratings-in-madden-nfl-13

------------------------------------------------------------------------------------------------------------

Today, EA released the first set of Madden 13 player ratings.  In this video:  http://www.youtube.com/watch?v=bkrJL1mMTHM&feature=youtu.be Hall of Fame running back Marshall Faulk revealed the top 10 QB's in Madden 13 - based on overall (OVR).  If you have followed me for any period of time, then you know I can't stand OVR or the current formula.  I have suggested for some time now that OVR needs to go way and EA should start using a stand alone ranking with no connection to attributes.  Attributes are consistently manipulated specifically to hold overalls down, despite statistics and tape that suggest the opposite.  The opposite happens as well, attributes are inflated to increase overall.  This can be seen in all positions throughout the game.  So what's wrong with that?  Well, by manipulating attributes to reach a targeted overall, the attributes tend to be inconsistent and inaccurate for all positions.  EA has conditioned fans to look at OVR first and this takes the focus off of the inconsistencies in the attributes themselves.  "Look beyond overall, it will open your eyes to how bad some of these ratings are."

EA basically uses OVR as a ranking system.  That being the case, they should just make it a stand alone ranking with no connection to attributes.  Let folks debate over who is #1 at each position, while assigning attributes based on statistics and on the field play.  Attributes should be based on reality, not popularity.  By having a stand alone ranking system, fans and players will have their bragging rights, but attributes will not have to be manipulated to change a ranking.  Each and every attribute should be rated as accurately and objectively as possible, regardless of OVR or RANK.

EA, stop handcuffing the player ratings staff and get rid of the overall (OVR) formula once and for all.


Ok, I will stop preaching now.  Let's look at some numbers.  I am going to take a 3 year look at each of these QB's and provide my take on this list.

Below are the top ten Madden 13 Quarterbacks based only on OVR (let's call it what it is, a ranking).

Best = The best in a category, out of these 10 QB's only.

Red = The worst in a category, out of these 10 QB's only.

SkFum% =  Fumbles caused by sacks, divided # of times sacked (fumble per sack ratio.)

RFum% = Rushing fumbles divided by the # of carries (fumble per attempt ratio.)


* P. Manning missed the entire 2011 season.
* M. Vick was not a starter in 2009.



I included Schaub, Ryan, Cutler, and Flacco for comparison.  Some people might think they belong in the top 10.

Observations:

1. Based on these statistics, there is a clear difference between the top 3 QB's and the rest of the group.  That said, why are Eli and Big Ben so close to the top 3?  Is it because they have two Super Bowl wins and the others don't.

BTW - In the last 3 years, no one in this group puts the ball on the ground more than Eli, plus he has a higher INT % then the rest of this group.  Winning the Super Bowl is fine and dandy, but should the ratings be based on the Super Bowl Eli or the regular season Eli.  Should a Madden player rating only consider the good times or should it also include the bad.  I believe it should include the good and the bad.  Finding the right balance is key to producing objective and accurate player ratings.

2.  Should one missed season really cost Peyton Manning that much?

3.  Doesn't Matt Schaub deserve to be in this group?  I think a strong argument could be made that Matt Schuab should be in the top 10 ahead of Michael Vick.

4.  Ok Cowboy fans.  Is Romo getting the shaft?  He does a great job at protecting the ball, completes passes at a very high rate, and had the 4th best QB rating the last 3 years.  Shouldn't the gap between Romo and Vick be more than one spot?

--------

* Some people don't put much stock into statistics, so lets look at this a different way.






* Stafford was select as a Pro Bowl Alternate in 2011.

Again, these figures are based ONLY on the last three seasons from 2009 to 2011.

Observations:

1.  The top three QB's have qualified for the playoffs each of the last three seasons.  They've gone to multiple Pro Bowls and have been selected to at least one All-Pro Team.  The only other All-Pro the last three years is Peyton Manning.

2.  While Eli is tied for the most playoff wins in the group, he only qualified to make the playoffs once in the last 3 seasons.  Of course, Super Bowl XLVI is still fresh in everyone's mind; some people may have forget the ups and downs that have defined much of  Eli's career.

3.  Outside of the top three, only Joe Flacco has made the playoffs each of the last three years.  Flacco doesn't get much credit for those playoff appearances, that typically goes to his dominate defense.

4.  During the playoffs, we witnessed some roller-coaster player attribute changes.  If playoff games give players a boost in attributes, then what about players who don't make the playoffs as often, like:  Romo, Rivers, and Vick.  I believe the larger body of work should have the most impact on player ratings.  It drives me nuts, when I hear people say the playoffs are what really matter.  This is the NFL not the NBA.  There are only 16 games in a regular season and every one of them matters.  When it comes to player attributes, one great catch during a prime-time playoff game should not be valued more then the same great catch in a locally televised week 7 game.  I know several people disagree with me on that, but to reach the goal of consistent player ratings/attributes, this needs to be the mindset of the folks on the Madden ratings team.  The "knee-jerk" style ratings system, does not build confidence in Madden fans (see:  Tim Tebow).
---------------

One more group of tables to look at.  The first table shows the difference between the final  Madden 12 OVR and the New Madden 13 OVR.   The second table shows the origial Madden 12 OVR compared to the new OVR.  I included the rankings as well.

Remember, my goal is to get rid of OVR and go to a stand alone ranking system that has no connection to player attributes.

Here are the tables:


Observations:

1.  This table really illustrates the hit Peyton Manning took for missing a season.  Remember, he was the Iron Man (208 consecutive starts) at the position before he got hurt.  Until we see all of the attributes, we can only speculate on which ones got decreased.  Since the Injury (INJ) attribute has no impact on the OVR formula, my guess it Donny Moore decreased both the Throwing Power (THP) and Deep Accuracy (DAC) of Manning.  There was a lot of talk about Manning's arm strength after the surgery and I believe Donny Moore jumped all over that.  Taking a wait and see approach may have been a better way to go, but I could be wrong.
--------------


Observations:

1.  I thought this was a very interesting table.  Look at the original OVR for these QB's in Madden 12 prior to any updates and compare that to their original Madden 13 OVR.  Stafford and Eli got some serious love over the last year, while Rivers, Peyton, and Vick took the most significant hits in the group.

2.  Based on previous seasons, there is a very good chance that 2 or 3 of these QB's will not be on this list by the end of the season.  A healthy Matt Schaub and a motivated Cam Newton are sure to make some noise.

---------


Thanks for following my blog, stayed tuned for more as the ratings continue to be released.

My August 5th blog on Madden 13 Quarterback SAC, MAC, DAC, CAR, and THP can be found here: http://maddenmanniac.blogspot.com/2012/03/attribute-spotlight-qb-deep-accuracy.html

UPDATE - Here is my newest blog on the top 10 Madden 13 Running Backs:  http://maddenmanniac.blogspot.com/2012/08/ea-madden-13-running-backs-my.html

Per Justin Dewiel, the full QB ratings will be revealed this Friday, followed by running backs next week.  WR's and TE's will be released in week 3 with defensive players rounding it out in week 4.  This is an exciting and frustrating time for me.  I will try to contain myself.  Have a great week:)

Here is the EA Sports link that breaks down when Madden 13 player ratings will be revealed:  http://www.easports.com/madden-nfl/news/article/ratings-reveal



Sunday, July 1, 2012

A look at Madden 12 QB Carry (CAR) attributes, plus NFL QB rushing and sack fumble %. (Madden 13 attribute suggestions included.)

This blog will cover both Madden 12 Quarterback carry (CAR) attributes and the actual NFL statistics regarding QB fumble percentages.  Be prepared for several tables and graphs full of data.  I felt it was necessary to include all possible data when looking at the consistency of this Madden attribute.  My goal is to improve the accuracy and consistency of the QB carry (CAR) attribute going into Madden 13.  I will include some projections at the end of the blog, that could be used for Madden 13 player ratings.

These statistics are widely available and can be found on various websites (Yahoo, NFL, ESPN, FoxSports, etc.).  I also used the official "NFL Guide for Statisticians" to help with some interpretation.  I decided to break the QB fumble percentage into two parts.  The first part was rushing fumble percentage and the second part was sack fumble percentage.  If you have followed my previous blogs, you know that no website provides actual fumble percentage per touch.  You have to take the available statistics and do the math yourself.  Finding the rushing fumble percentage was pretty easy. I divided total rushing fumbles by total attempts to get a per touch rushing fumble percentage for QB's.

Now finding a sack fumble percentage for QB's is not so easy.  Several websites provide rushing fumbles for QB's and total fumbles, but they don't specifically indicate sack fumbles.  I had to determine what was and wasn't recorded as a rushing attempt.

I will try to summarize.  The most important part is that a QB does not get charged with a rushing attempt on a sack fumble because the QB is in the act of passing and not rushing the football.  So, if you subtract the rushing fumbles for a QB from his total fumbles, the result will be fumbles that occurred due to being sacked.  This was key for me to determined sack fumble percentage.

What about fumbled snaps?  Unless it's determined that the Center made a bad snap, the QB will be charged with the fumble.  The QB is also charged with a rushing attempt and fumble during a muffed hand-off.  It could be argued that the RB is also responsible for a muffed hand-off, but the NFL charges the fumble to the QB.

I expect some fans to disagree with my logic when looking at the QB Carry (CAR) attribute in Madden NFL football, but I believe it's the best available data.  As it stands right now, EA will not share how they assign Madden ratings or attributes.  After reading this blog, I hope you will see that having established guidelines for attribute ratings is long over due.
---

First let's take a quick look at the fumble percentage of other positions during the 2011 season:

Total touches = rushing attempts + receptions for RB, TE, and WR.

Total touches = return attempts for PR and KR.

FUM% = Fumbles/Total Touches



After looking at this chart, you can see that RB's, WR's, and TE's were very similar when it comes to fumble percentage.  Based on this chart, the player's with the highest risk to fumble are the Kick and Punt Returners, especially the Punt Returners.  If you think being a Punt Returner is high risk, wait until you see the QB statistics I'm about to show you.
---

Here is the combined rushing and sack fumble percentages for all 2011 NFL Quarterbacks (regular season):

Total FUM % = All rushing attempts + Sacks / Total fumbles.


I knew it was bad for QB's, but those numbers are staggering.  The average QB fumbled six times more than a RB per rushing attempt.  For every 6 sacks, the average 2011 QB would fumble the ball.  The total Fumble percentage of 10.25% blows away all other NFL positions.

---

I collected the CAR attributes for all REAL NFL QB's on the final Madden 12 rosters.  I say real, because there are some developers that have characters in Madden, so I excluded those.  Here are the results:

Notice the average carry (CAR) attribute for QB's is 62 and both 65 and 60 are the most assigned CAR attributes.  What you will see as you read through this blog, is that several QB's are actually well above average at protecting the football based on NFL statistics, but are below average in their Madden CAR attribute.
---

Next I gathered the career statistics on 76 QB's who were active during the 2011 NFL season.  I compared this data with their assigned Madden 12 carry (CAR) attribute.  All CAR attributes are from the final roster update for Madden 12.

Blue = Below the Average for these 76 QB's (The lower the better.)

Red = Above the Average for these 76 QB's


It should not take long to realize the career fumble statistics for most of these QB's were not considered when assigning their Madden 12 carry attribute.  The numbers speak for themselves, the QB carry (CAR) attribute is terribly inconsistent and flat-out inaccurate for the majority of these QB's.

---
Just for fun, I gathered the statistics on some retired QB's.  Some Hall of Famers, and some draft busts.  Here they are:


Blue = Below the Average for the 76 active QB's (The lower the better.)

Red = Above the Average for the 76 active QB's



Due to how fumbles were tracked for QB's, I could only go back to 1991 for Young, Cunningham, and Elway.  That said, I still think it's a good enough sample size for comparison.  It shouldn't surprise you that Steve Young was great and Russell was horrible.  I think the fact the Kurt Warner was so bad at protecting the ball will surprise some folks, unless you are a Rams or Cardinal fan.
---


So, I got to this point with a ton of questions.  I decided to break out my scatter plots to answer them.






As discussed in my previous blogs, any R^2 value over 0.80 is considered a strong correlation.  The closer the R^2 value is to 1.0 or -1.0 the stronger the correlation.  If you are interested, here is a link that explains this type of statistical analysis:  http://mathbits.com/MathBits/TISection/Statistics2/correlation.htm

Questions:

1.  Is a QB's sack fumble percentage higher if they are sacked more than other QB's?  ANSWER = NO.  The graph in the top left has an R^2 value of 0.0036.  Based on that number, there is absolutely no correlation between the number of times a QB is sacked and their sack fumble percentage.  We must remember, it's NOT the total number of fumbles that matter here.  What matters is how often a QB fumbles when sacked or rushing with the football.  The per touch statistic will identify who is actually better at protecting the ball and who isn't.

2.  Does the number of rushing attempts effect how often a QB will fumble per touch?  ANSWER = NO.  This time, the graph in the top right indicates that the two values are almost completely uncorrelated (R^2 value of 0.0323).

3.  What if you combine total rushing attempts and total sacks?  Does that impact the total fumble percentage?  ANSWER = NO.  The bottom middle graph has an R^2 value of 0.016.

So what do those graphs tell us?  It's pretty simple really.  Some QB's are better at protecting the football than others, regardless of the number of rushing attempts or sacks taken.  Don't confuse playing time with how often a player will fumble.  A 6 year veteran QB will likely have more fumbles than a 2nd year QB, but he could fumble at a much lower frequency.  The carry (CAR) attribute in Madden helps determine the frequency that a player will fumble.  That is why using a fumble percentage make the most sense, it tells you the frequency based on the actually NFL statistics.
---

I had a couple more questions to address, so I did three more graphs based on the statistics from the 76 QB from 2011.  Here they are:




1.  Does EA use Career Fumble totals when rating the carry (CAR) attribute for QB's?  ANSWER:  NO.  Believe it, or not.  This is the answer you want to see.  It's not about the total number of fumbles.  It should be about the frequency of the fumbles.

2.  Does EA use fumble percentages when rating the carry (CAR) attribute for QB's?  ANSWER:  NO.  None of this these graphs indicate that EA uses fumble percentage (frequency statistics) in anyway.  In fact, these R values are horrible and show nothing more than inconsistency.
---

What about the 2011 season?  Maybe EA adjusted throughout the season and it's not based on the career percentages.  Here is a sample group of 39 QB's from 2011 and their final Madden 12 carry attributes:



By just looking at the table, it doesn't appear like EA used the 2011 season statistics to more accurately assign the CAR attribute as the season progressed.  Personally, I think you need to look at more than one season to provide the most accurate and consistent ratings.

Here is the graph for these 39 QB's from 2011:


This graph offers the same results as the previous graphs.  Remember, the closer the R value is to 1.0, the stronger the correlation.  The R value of 0.1119 show the exact opposite, little to no correlation between M12 CAR attributes and the 2011 fumble percentages.
---

One final question came to mind.  Did EA use the 2010 regular season fumble statistics for Madden 12?  It would make sense that the previous season would be looked at when assigning player attributes.  I selected a sample of 29 QB's who played the most during the 2010 NFL season, then I charted the statistics and did a graph.  Here are the results:




Once again, there is little to no connection when looking at the 29 most utilized QB's from the 2010 season.
---
So what does EA use to assign the carry (CAR) attribute in Madden football?  ANSWER:  I have no idea.  I think they do a lot of guesswork and reacting to a fumble during a big game.

Did you know, the CAR attribute has absolutely no effect on a QB's overall (OVR) rating in Madden 12?  Since it has no effect on a QB's OVR, it should be even easier for EA to assign accurate and consistent ratings.  I have said several times, that I wish OVR had no connection to any attributes and was a stand alone ranking only.  In this case, I get my wish (CAR has no connection to QB OVR), but fans still have to deal with inaccurate/inconsistent attributes.

Several people like to scramble with QB's in Madden football, but they pay the price with fumbles.  I've had many people tell me that Vick and Rodgers fumble too much in Madden; based on these ratings, they might be right.  In general, many QB's are not very good at protecting the ball when rushing or during a sack, but some are clearly better than others and that should be reflected in Madden gameplay.
---

I can't end this blog without giving some suggestions for Madden 13 player ratings.  Below I will include my first attempt at assigning QB carry (CAR) attributes based on career rushing and sack fumble percentages.

This is not where I would stop.  This is only the first step in assigning ratings.  No one person should make all the decisions.  


I have been asking EA to expand the Madden player ratings team.  The team is necessary to help fine tune the final attribute ratings.  For instance, I would suggest discussing a player like Hanie who has limited NFL playing time and statistics.  Maybe a minimum number of games played should be required to reach the elite level of an attribute rating.  I am not saying Hanie should have a higher CAR attribute then Cam Newton, this is just the starting point.  The next step is discussion with the ratings team.  Unfortunately, EA's resources appear to be limited when it comes to player ratings.

In my opinion, a chart like this should be the first step for assigning Madden NFL 13 Quarterback carry (CAR) attributes:

M12 CAR = Final Roster Update carry (CAR) attribute in Madden 12.


The carry (CAR) average for all Madden 12 quarterbacks was 62, my average dropped just one to 61.  All of these suggestions are up for debate, but I believe this to be a critical and reliable first step to take when assigning these ratings.

You can see that I used the career statistics as a starting point.  I wouldn't be against looking at the last 3 years for veteran QB's to see if they have improved or declined when it comes to fumble %.  Again, this is something a ratings team could decide and fine tune.

I also choose to use the Total Fumble % when assigning my suggested ratings.  It would be awesome if Madden could have both a rushing carry attribute and an attribute that determines the frequency of sack fumbles.  Of course, the defenders play a role when trying to strip the ball, but the QB is more responsible for the fumble in my opinion.  Some QB's hang onto the ball too long, others don't carry the ball in a protected way, and others just don't sense pressure very well.  Those things are all on the QB in my opinion.

Some people might think the CAR attribute isn't important.  But if you've lost a game to a fumble or several fumbles with a QB who should have had a higher CAR attribute, then you have a right to be upset.

On the flip side, the perception is that a player like Tim Tebow is invincible when running with the football.  While he is above average for a QB, he should not be #1 with an 88 CAR, especially if the CAR attribute contributes to the frequency of fumbles in Madden.
---

Thanks for following my blog.  If you missed my Running Back carry (attribute) blog, you can find it here:  http://maddenmanniac.blogspot.com/2012/02/attribute-spotlight-running-back-carry.html

Have a safe 4th of July and remember, DON'T DRINK AND DRIVE.

Follow me on twitter @mannmicj