Experimadden: The Flats & CPU Defensive Adjustments
This is going to be a bit of a long post, but I want to explain myself fully to make everything clear. Bear with!
Hello all,
I picked the game up nearly 48 hours ago now. I was planning on participating in three different leagues: one offline, one online with 10 players, and one full all-human league. So, I decided to play a TON of exhibition games with various teams and settings, to get a feel for the game, the teams, the playbooks...you know the drill.
Over the past two days, I've noticed that it's incredibly difficult for me to have much success throwing down field. Running has also been spotty. In contrast, I've spent increasingly more time making plays in the short passing game - passes to running backs and tight ends no more than five or six yards down field. Lateral plays.
The flats.
This pattern developed naturally. In one of the last exhibition games I played today (With Houston) I realized how insane things were getting. Even with a variant of Marino's sliders, I was consistently having about 20 receptions combined between Owen Daniels and Steve Slaton, without much coverage...ever. Even with reasonably tough passing sliders, and a decent AI reaction time.
I came to the forums and realized some other people were upset about the flats being easy access. So, I decided to do an experiment. Here was the setup:
- Seven Minute Quarters (No accelerated clock)
- Custom Sliders (Based off of Marino's sliders found here on OS)
- I played as the Houston Texans against the rival Jacksonville Jaguars, in Reliant Stadium
- All defensive snaps via supersim (Mainly to save time - I only cared about the offensive side of things)
And here's the "Experiment" part:
The only play I ran on offense the ENTIRE GAME was to be a pass play to my running back in the flat, out of one of Houston's shotgun sets. I don't remember which set - don't have it in front of me and forgot to screenshot it - but you can see the shape of the formation in one of the later screenshots to figure that part out. The play is called "Double Drags". I did not allow myself to audible. I did not allow myself to scramble. The only thing I allowed myself to do was to flip the play to allow it to always develop to the open side of the field (I alternated hash marks every other play due to the directional nature of things). Fatigue was on, so Slaton was supplemented by Ryan Moats for a number of touches.
Here are a few screenshots showing the relevant difficulty settings:
Player Skill:
http://i30.tinypic.com/2hrlgy9.jpg
Passing Sliders:
http://i27.tinypic.com/2v193qa.jpg
Pass Defense Sliders:
http://i31.tinypic.com/28wpdu9.jpg
I ran the play 58 times. Here is a screenshot of my recent play tab showing ONLY Double Drags with 25 seconds left (I scored once after this as well)
Recent Plays:
http://i30.tinypic.com/2hn6z38.jpg
Here are the end results...
Final Score:
http://i32.tinypic.com/hsm7o5.jpg
Schaub Att/Yards:
http://i29.tinypic.com/b4sw1x.jpg
Schaub Comp/TD/INT:
http://i25.tinypic.com/2vdiu8o.jpg
RB Names/YAC:
http://i31.tinypic.com/m7q71i.jpg
RB Rec/Yards/AVG/TD:
http://i30.tinypic.com/jv6a6g.jpg
---
There are a few things I want to point out. Firstly, I didn't plan on simming a defensive goose egg on this one. Apparently the Jaguars were so deflated by the Slaton/Moats onslaught that they just decided not to bother. Also, the third running back randomly was in for two plays in the third quarter - otherwise Moats and Slaton switched back and forth depending upon fatigue. I made no manual subs at any point. I also (obviously) ONLY threw to that running back - whoever it was at the time.
I want to stress that I'm not a Madden *****. I've been excited about this game for months, and I still appreciate the hard work that went into it. It shows, even with the glaring flaws. However, I wanted to see just how badly the AI would handle this experiment...and to say that it failed would be a colossal understatement. And here's something scary:
I did this with "Sim" sliders. Imagine how much worse it would have been on Pro or All-Pro? This is the kind of ridiculousness that should only be possible on Rookie. Outside of flipping the play, I NEVER changed my pattern. EVER. If the defense called an appropriate scheme, it was at random.
Anyone who points at my use of Slaton and tries to say that my results are largely due to having a great Running Back to play with - compare the numbers between Slaton and Moats here. There's a little drop off, but in essence they both easily averaged a first down. I ran this play 58 times in a row with Sim sliders (Good sliders, even) and averaged a first down.
I threw no interceptions.
Two passes were tipped.
Now, this isn't a money play. My criteria when picking a play to do this with was (1) Is it in the shotgun? and (2) Is there a RB to pass to in the flat.
That was it. I could have generated similar results with any RB pass to the flat. I also believe I could do something similar with TE out routes and similar short passes.
What's the point?
The point here is twofold. Firstly, the flats are broken. Even when Slaton was in man coverage, his defender was incredibly far away. In zone coverage, at best the defender hit the receiver near the point of contact for a very short gain - but even then, I was often able to break down field with a simple juke or a spin move. The flats are broken. The entire underneath passing game is broken. I won't be testing this online, as I don't WANT to prove that it works online if it possibly could. Short of imposing ridiculously expansive house rules on myself, there's no way for me to enjoy a game against the CPU at this point. It would be one thing to avoid calling one play - but to avoid an entire underneath passing game because it's not being defended correctly...ever? That's not fun at all.
The second part to the point here is that the defensive AI and playcalling is ABSOLUTE RUBBISH. It's 2009. Madden has been around 21 years. It's time that cpu controlled defenses actually pay attention to what I'm doing. All the defense was doing was analyzing the down and situation. At no point did the defense adjust. That running back was never double covered. They continued to send the same CB blitz intermittently despite the fact that it NEVER worked, and the fact that I ALWAYS gained an extra 10 yards because of the vacated area.
This has to change. I know that it's difficult to develop a real functional AI for a football game - baseball is comparatively easy - but if I play an exhibition game on MLB: The Show 09 and throw every single fastball down and in with the same velocity...the game catches on. If I drill the changeup out of the park twice...they stop throwing the changeup.
Madden 10 is an improvement in many ways. However, this is simply broken. The out routes have always been a bit too money in Madden, but now the flats are broken too? How long until the top of every leaderboard is occupied by players who have figured out how to translate this strategy to an online game? It's frustrating.
I did this to expose the problem. I'm not trying to rub EA's nose in it. I want this game to be great. I want to be able to play an OFFLINE franchise and feel like I'm genuinely challenged by the game. I don't want to have to impose five house rules on myself to even make the game challenging to begin with. Sometimes, I want the game to outplay me. Outsmart me. To beat me.
This isn't going to cut it. I have no idea what to do now - I'm never going to feel like I can pass to the flats again in this game without feeling dirty afterward. I might just do one sim-based franchise to completion and sell it, honestly.
Now you know. Feel free to try and replicate the experiment with the same play/team/playbook/opponent, or with alterations. I promise you it went as described here. I'm curious how ridiculous the numbers would have been on Pro.