Home
Reply
 
Thread Tools
Old 07-31-2013, 03:25 PM   #1
MVP
 
infemous's Arena
 
OVR: 6
Join Date: Nov 2009
Blog Entries: 1
How to Improve CPU CFM AI.

I think we all understand that Madden's CFM CPU AI is lacking, which can lead to some undesirable roster decisions that can really stifle our enjoyment of the mode.

I think everyone is sick and tired of seeing teams with a franchise QB drafting a rookie QB 1st Overall, or re-signing a 33 year old RB to a 4 year contract, or letting valuable special teamers such as Trindon Holliday walk and never get signed because he isn't a good WR.

There are a couple of steps that must be taken by the developers over at EA Tiburon before this will be a completely enjoyable mode, although I will do my best to avoid descending this post into a wishlist. Things I am sure people would love to see (such as interviews, added draft presentation, combines, weekly shows etc.) will not make this post, but a couple of necessary steps for the progression of CFM which aren't currently in the mode will.


1.

All ratings must be clearly delineated and a logical weighting for OVRs, or the abolition of OVRs must take place.


Nothing is more frustrating than seeing a team draft a reasonable speed but excellent route running WR high in the first round, only for his poor speed or low CIT or whatever weighs heaviest into his OVR assigning him to a low position in the depth chart, and when he sees the field, he gets locked down by a 60 OVR CB with 99 speed.
If individual ratings matter, and all this means ALL of them, not just a select few, then roster decisions will be informed by a gameplay representation, both for the CPU and the human player.
Remove things such as elusiveness that render trucking and juke move ratings useless, make route running count, make awareness for QBs either count for human players or make it less valuable in the OVR. Pass Blocking needs to determine the ability for an O-Lineman to hold a pass block, not his strength rating, this will allow for schematic differences where zone blocking OL can be useful when in the correct scheme, while power blocking schemes will have more of a need for powerful OL.


Additionally, consistency and injury ratings should play a huge impact into OVR as well as any traits they have, should the dependency on OVR for CPU logic remain. What makes one 85 OVR QB better than the other? Things such as being clutch, consistent, throwing a tight spiral, their poise in the pocket, etc. etc. If the CPU cannot simply understand that these traits are also important, make it obvious to them by putting it into the OVR formula.
Additionally, pass protection for RBs, run blocking for WRs etc. must also be included into the OVR formula.

2.

Revamp depth charts and open up player positions.


I touched on it in 1, but seeing players drafted high, only to be cast under and buried on depth charts for the rest of their careers, never seeing the field and giving them an opportunity to progress, is a huge issue that would be remedied by including specific positions on depth charts for players with specific skill sets.
There is currently a depth chart for a 3rd Down RB, yet there isn't one for a goal-line RB, #1 WR, #2 WR, slot WR, nickel CB, dime CB, situational edge rusher, goal-line defense, punt block team, kick return TEAM (not just returner), kick coverage/gunners etc. etc.

Having an in depth depth chart would allow for teams to weight players' unique abilities into their roster decisions, with teams' propensity for 2 TE vs 3 WR sets to be dictated by the ability of their personnel, or the importance of the 2nd TE vs the Slot WR to be weighted by the teams' playbook/scheme.

Additionally, this would really help with the setting of XP targets and goals for players, as they would be expected to attain certain numbers depending on the importance of the position they are placed into on the depth chart. A slot WR for example may be expected to catch more balls than a number 2 WR, but a number 2 WR may be expected to catch more TDs. Likewise, on defense, a dime CB will not be expected to make as many interceptions as the number 1 CB, but his XP gains will be directly influenced by more general team goals or the number of snaps he sees. (A player should get XP for seeing the field, even if it is minute, this would promote the use of different packages and personnel groupings both for humans and CPU teams) Any failure to achieve the goals should be more punished however, to avoid rampant improvement and to balance the progression/regression elements. I believe I have made a post on this site somewhere that goes into more depth about how I feel the progression and regression system can be improved so I will stop here in this post.

In order to assist the CPU into being able to find out which player is best suited to which position, a player must not be constrained to one specific position on the field. A WR with ability to play multiple spots, such as Slot, Outside, RB or even Wildcat QB should be listed as having these spots as secondary, tertiary etc. positions. This would be especially helpful for the O-Line, in which interior OL could be played across the line depending on injuries and ability, while also clearly defining a 'true' OT or one that can be kicked inside should a better player be signed at their position.

This would also help unclutter playbooks, which are currently littered by peculiar personnel groupings and subsequently constantly changing buttons for players in the same spots on the field. I don't know if I am alone in this, but having my slot WR have his button change from L1 to triangle to square and whatever else it changes to, is incredibly frustrating and can lead to unnecessary interceptions, which happen far too frequently because of CPU cheese anyway lol
Now playbooks will have packages for certain situations, such as 3rd and long that would bring out the best pass rushing package, as assigned by the deeper depth chart system.
This will also be incredibly helpful in the moments when a player is injured for a quarter or two during a game and your depth chart is royally messed up until the players' return.

3.

The return of Player Roles.


Personally I have no idea how on earth this addition got removed from Madden, but it is a really sore point for me. Having a 1st round tag on a player helps the CPU understand that the player has been invested in and should not be immediately traded (it should also let a team know that they should actually play that player... but that should hopefully be addressed by points one and two, with this one making sure of that lol)
Additionally, having a Franchise QB tag on a player helps make sure that a team puts its best effort into keeping that player when it comes to re-signing them. I found that these roles had the biggest impact when it came to draft strategies and that is something that should not be neglected.
Again, I feel that these Player Roles should also be weighted into a players' OVR, or at least have an impact in a players' position when it comes to negotiating new contracts.

Teams should also have 'roles' that are retrospectively added to them, helping indicate whether or not they are a superbowl calibre team, playoff contender, in a rebuilding mode, emerging, old/past their prime, young, etc. etc. which will help inform off season decisions. Does a team invest heavily into an aging veteran in the hope it pushes them over the edge into a Superbowl calibre team? Or do they sign the depth and role players that will assist their already superbowl calibre team? Or sign depth to give the rookies they've invested in time on the field to shine?

4.

Morale and Injury System.


As a progression and added depth to the player roles; team, player, coach and owner morale should play a huge part.
It can start off as a very simple system and gradually get more in depth as new maddens come out.
If a player is playing, the team is winning, his contract reflects his production and he is healthy, his morale should be good. Any of these things aren't happening, and the player's morale is likely to sour. He won't want to re-sign with a losing team (this will be very much weighted by age. A veteran would rather take smaller bucks and join/stay with a superbowl calibre team than a young player who is looking to get paid in his second contract and will thus be more open to joining a rebuilding team etc.)
Likewise, a coach can over time, after consistently losing, lower his own happiness, and will subsequently not be as effective (with regards to his perks or abilities) or not want to re-sign with the team when his contract expires. He may rather retire or take a step down and become a co-ordinator.
If an owner is unhappy, typically people get fired. An owner's happiness would be weighted by their expectations and the results of the team. Things such as the last 3 years' results would be taken into account too. If a rookie coach comes in, an owner may not be expecting much. If the rookie coach only musters a 3 or 4 win season, he may not be pleased, but he'll abide it. If after 3 years the owner can see a continued gradual improvement from 3/4 wins to 5/6 to 7/8, he is less likely to fire a coach and may give him a prove it deal (one year extension or honour the last year of the contract) with the hope of more improvement. Likewise, if a veteran coach goes 12-4 in one year, then falls to 1-15, the owner may give him another year... Should there be an improvement or return to form, the coach will stay, but if the team follows up with another stinker of a year, that coach will be gone.

The league is inhabited by people, not algorithms and numbers, and it is important that the Madden tries to emulate this. CFM mode needs to take into account the personalities of every person involved in the league and their situations, so as to achieve a logical or understandably illogical conclusion.
The fun for the developers will be the creation of anomalies to conventional wisdom, the Al Davis type owner, the Percy Harvins, etc.

The injury system will be integral to morale too.
Firstly, there needs to be a wear and tear injury system in game. Once that is established, it will need to be included in morale and roster decisions.
If a star player gets hurt, his morale will be lower. The player's ability to rebound from that injury, or the team's faith in the player to do so, will greatly impact the next contract they give him, whether they cut him or whether the team progresses or regresses. There are numerous real life examples, but the Peyton Manning situation is one that would be reflected through this inclusion. Likewise the Jake Long situation, and his eventual contract, as opposed to the Adrian Peterson situation. If a player has a good season after being injured, his morale when injured will be less impacted, allowing for the player and the team to have more confidence in the player. This would be countered and tempered by the team having to be wary of an injury getting worse and eventually ending that players' career.

A wear and tear injury system would play a huge role in the regression side of the game, which will obviously inform roster decisions.

5.

If its in the league, its in the game.


PUP list, waiver wires, holdouts, Restricted free agency, practice squads, conditional draft picks etc. etc. must all be in the game.
Give the CPU and the Human every possible alley to improve their team and help manage their finances.
By including all these things you are recreating the small but integral elements of the NFL that make it so.
Player gets hurt in training camp, or their recovery is slower than expected (as injuries will obviously carry through from season to season)? Teams can put them on the PUP list and hope for their return.
Additionally, teams will have to declare who will be active on gameday too, with injury reports being in depth and being a vital element to not only offseason roster management but gameday roster management too.

6.

Media presence.


This could be continuation of both the Player/Team Roles and the Morale system, but essentially, make the media a presence in the game.
Interview coaches or star players, award winners or underperformers during the season. Answering in specific ways will have notable (but not exaggerated) implications for morale, the way the team values the player, the way a player values the team etc. etc.

An example: Star WR for the Rams gets hurt.. The media ask Jeff Fisher how the team will cope without him.
If he says the team won't cope, the other WRs will not be happy about it.
If he says the team won't miss him, the injured star WR may get annoyed.
If he says the team is full of guys able to step up, the players may feel neutral but the fans could either place too many expectations on a player who isn't able to step up, subsequently effecting their happiness when he doesn't, or the new player may step up and push the star, now injured WR off the team.

Every answer or refusal to answer will have a possible positive and possible negative outcome.
It seems this is making its way into Owner Mode in M25 and it is essential this is applied across CFM as a mode, especially for Superstar and Coach mode.

These adjustments and added nuance to the mode will help ground any CPU decision in something understandable for the player. A big star player being cut will be less baffling or ridiculous if you saw him openly criticise his coach after a bad loss in a rivalry game or he was ousted by an emergent talent.

7.

Free Agency and re-signings.


Every contract in the game must be negotiated.
The logic must be tuned using all the variable elements mentioned above, however, if the Free Agency system remains the same, all these additions will hold virtually no weight to the real crux of CFM.

You cannot go into the Free Agency pool in the real NFL and offer a player a league minimum 1 year deal and know he will sign with you.
The current Free Agency system in the game is excellent, albeit flawed, but should simply be applied throughout the season.
Every time you want to sign a player, you have to offer him a contract and negotiate it with him. The player will be able to sit on your offer for a while and see if another situation suits him better, he will be able to sign it immediately and get to work, or he could decline it and wait it out.
The current system is undermined because after the draft, that 89 OVR QB who rejected every single contract offer during the free agency period, is now fair game to whoever has the cap space to sign him immediately. Why would he not sign before the draft? If that is informed by a logic, then it should inform the logic he will use when it comes to signing with a team after the draft n'est pas?

8.

Coaching Carousel


All player logic and player re-signings should be equally applied to coaches.
There must be full rosters of staff, all licensed, both in the league currently or free agents, and if that isn't possible, we should be able to create as many as we want.
Each coach will be assigned a playbook, scheme and player preference, while also having a reputation which will work (better than but) similarly to the legacy score. If an OC is in charge of an offense that consistently ranks in the top 5 in the league his reputation will increase and he is gonna get a HC shot soon.

Teams must be able to interview/negotiate contracts with free agent coaches, with the coach's morale or scheme, etc. playing a huge role in the position they take, as well as how much money is earned. Building a staff would be integral to player progression, with each coach having one start up perk, such as a coach having a high 'likeability factor' that would help improve morale, or an XP boost for specific positions they excel in etc. This is touched on in the current CFM, but will obviously be way more in depth with the aforementioned additions in this post, and will be applied to every single staff member in the game.

9.

Contracts and extensions/restructures.


Contracts must be able to be weighted depending on each specific year. You can pro rate a bonus, backload or front load a contract, pick a specific year to keep a cap hit especially low or pick one year to stack it. They must include bonuses (which will affect morale) and player logic when signing them must be improved.

Offering a 36 year old RB a 5 year deal and back loading his contract would not go down well in reality so it shouldn't in Madden.

Player contracts should be able to be restructured, depending on the player's morale. If you want to restructure a player with low morale's contract, he is likely to tell you to clear off and let him hit the market, but a high morale, franchise player such as Tom Brady is much more likely to negotiate a cap friendly deal that redistributes the money he is already owed.

You should also be able to extend a player's contract at any point, and if you fail initially, that should not mean the player will never speak to you again. You (and the CPU) could employ the media to boost the player's morale and bring him back to the table... but beware of alienating other players when doing so.

10.

Deeper statistics.


Advanced metrics are currently en vogue right now, with sites such as Football Outsiders and Pro Football Focus being used not only by fans looking for more in depth information about the league and its players, but also teams.
Madden is a computer game and records every single outcome, knows who was the intended target on ever play and who's coverage a pass went into.
I know that the assisted tackles thing has been an issue for everyone for a long time, however if they are able to do points 1 through 9, they should be able to fix and expand upon counting assisted tackles.

Knowing a player's catch %, a defender's coverage % (or grade), the way a rotational DL influences a run play without making the tackle, which OL is allowing the most pressure, the true yards after contact of a RB, etc. etc. Would revolutionise how teams assess their talent and allows for the role player to be recognised by the CPU beyond their OVR rating or fit within the new depth chart system.
This will also be huge for differentiating what teams focus on or look for in free agents, as well as progression and regression (more in depth goals than pure yards or TDs, INTs etc. etc.) and could really help the CPU apply the Player Roles to players. Darrelle Revis has a 3% completion percentage over the last 3 years and has only been targeted 50 times despite playing every snap? He is a certified shutdown CB.
A DT has recorded 1 hit and 2 hurries over 3 years despite playing 50% of his team's snaps but then has recorded 30 TFLs? He is a run stopping behemoth ONLY.

This will add not only depth and a more accurate application of player roles, but will cater to every player who likes statistics, while adding a scouting element into the game, which will greatly assist Online Career Mode players.
__________________________________________________ ______________

This is all I can come up with for now, but I am sure there are many things I missed.

The mode would be improved greatly, not only for the CPU AI, but for the human player's ability to work within an NFL world, and putting the CPU and the Human player on an equal level with regards to roster management.

Obviously, with EA's track record, these features would all have to be implemented incrementally, but these are all feasible additions (or returns) that have been done in other games or in previous Maddens.

All these additions would signify a commitment to a recreation of the NFL, and return unfathomable amounts of faith back into EA.
With this promise, I offer these ideas to you free, EA Tiburon, and hope you implement them.

Thank you very much for reading, I apologise for its length, but I hope you enjoyed it and I look forward to the discussion about it that can take place

Whether you are a gameplay, presentation or Career mode guy, I find it hard to believe anyone would be opposed to these additions, and I look forward to reading your additional suggestions or improvements too!


Last edited by infemous; 08-01-2013 at 04:46 AM.
infemous is offline  
Reply With Quote
Advertisements - Register to remove
Old 07-31-2013, 07:21 PM   #2
Rookie
 
ohmyJEBUS's Arena
 
OVR: 0
Join Date: Jun 2013
Location: New York
Re: How to Improve CPU CFM AI.

I've always felt as if madden was such a "bare bones" game for a game that is produced by such a large company. I strongly recommend posting this on an EA forum, or somehow putting someone from EA to this thread.

Good Stuff
ohmyJEBUS is offline  
Reply With Quote
Old 08-01-2013, 12:35 AM   #3
Rookie
 
OVR: 3
Join Date: Feb 2010
Re: How to Improve CPU CFM AI.

Awesome stuff. I'd have a hard time putting that game down.
rootofalleli is offline  
Reply With Quote
Old 08-02-2013, 04:43 AM   #4
MVP
 
infemous's Arena
 
OVR: 6
Join Date: Nov 2009
Blog Entries: 1
Re: How to Improve CPU CFM AI.

Thank you!

I added a 10th point because I remembered something that was crucial to the idea, but didn't have time to do yesterday.

Hope you enjoy the post and please, if anything can be expanded upon or should be changed, let me know and I'll update it some.
infemous is offline  
Reply With Quote
Old 08-02-2013, 09:22 AM   #5
Rookie
 
OVR: 4
Join Date: Nov 2009
Location: Palm Beach,FL
Re: How to Improve CPU CFM AI.

Great points.
I really wish they would add this stuff but I don't think they will.
Madden is afraid of turning off the casuals.
LaW97 is offline  
Reply With Quote
Advertisements - Register to remove
Old 08-03-2013, 07:04 AM   #6
MVP
 
infemous's Arena
 
OVR: 6
Join Date: Nov 2009
Blog Entries: 1
Re: How to Improve CPU CFM AI.

I don't see how this would turn off casuals when casuals play mostly online head to head?

That sort of thinking by EA has been the issue all along.
infemous is offline  
Reply With Quote
Old 08-03-2013, 06:05 PM   #7
Pro
 
Robo COP's Arena
 
OVR: 3
Join Date: Feb 2012
Re: How to Improve CPU CFM AI.

I didn't read this yet, but wanted to chime in on my dislike of the CPU AI as well. Completely ruined the game in madden 13. I love building a bad team into a good one over a couple years with madden but when franchise QBs are hitting free agency year after year it's a done deal. I could justify here and there a QB leaving (pretend like it wasn't enough money, bad team, etc.) but when RGII and Andrew Luck are 90 ovrs coming off of their rookie contract (and RGIII just came off of a SB win) and they go to a different team in free agency, THEN you know the AI logic is severly lacking.

I'm sure you have a lot of great points, unfortunately EAs fix is allowing the player to control all 32 teams. Rather than fix the problem they'll make it so the player has to fix the problem themselves. Hopefully they do extensive work on the AI in next gen.
Robo COP is offline  
Reply With Quote
Old 08-03-2013, 06:34 PM   #8
Hall Of Fame
 
KBLover's Arena
 
OVR: 40
Join Date: Aug 2009
Blog Entries: 14
Re: How to Improve CPU CFM AI.

Just awesome write-up.

Gosh I wish we could get together and make a game.
__________________
"Some people call it butterflies, but to him, it probably feels like pterodactyls in his stomach." --Plesac in MLB18
KBLover is offline  
Reply With Quote
Reply


« Previous Thread | Next Thread »

« Operation Sports Forums > Football > Madden NFL Football > Madden NFL Last Gen »


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is On
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -4. The time now is 08:15 PM.

Top -