League Forums

I have completed the first in-depth analysis of all the plays in the playbook. Basically I took two teams, assigned everyone to have 90-rated attributes across the board, and ran multiple simulations of each offensive play against each defensive play, in order to identify if there are any plays that are unstoppable or any plays that are completely useless. And to my surprise, it does look like there are generally successful offensive plays for each defensive play and vice versa.

So here's something I'm thinking about building out and trying in place of the current overuse penalty:

When you scout your opponent, the plays you select will also tell you the top 5 plays to oppose that play (based on this data, not your team's strengths vs. their strengths). You will then have a control to audible into one of those plays if you see your opponent line up in that formation. The controls I'm thinking of are:

* Seen the scouted play X times in the game (can be zero, but you probably would want at least 1 just to make sure your opponent is actually using that play in their gameplan)
* Restrict to time/quarter/score in the same way as rules
* You would set a weight for each of the 5 best plays against the given opposing play
* You are responding to your opponent's -formation-, so this logic would happen after both teams have selected their play. The offensive would decide if they want to audible out, then the defense would respond to the offense.
* For simplicity, I will probably fudge on the personnel (so the play you audible to doesn't have to match the personnel of the play you are audibling out of, and the players on the field would match the play you are audibling to instead of requiring you keep the same players on the field after you've changed plays)
* The play you are using as a base for audibling against must be in your scouted plays, and the plays you audible into must be in your game plan.

Thoughts or comments?

In the interest of being thorough, I should point out there are two plays that consistently were neither big gainers or big losers. Those were the spike play and the victory formation. All other offensive plays had a positive average.

Were there any defensive plays that had a negative average? Negative meaning the offense consistently lost yardage. Conversely, from the defensive perspective, were there any defensive plays that had more than a 3 yard average on yards gained by the offense?

Every defensive play had a positive average yards gained against across all offensive plays - even the multi-blitzes. Every play had a negative average against some other plays, but when taken across the entire spectrum of the playbook all plays averaged positive.

Remember also that this test is with all equal players on the field. It's possible - likely, even - that the audible system will fail for a given team because they just don't have good enough players to overcome their opponent. But the idea here is not to make a great opponent bad, but to give a better counter to teams overusing a specific play or set of plays.

I like this as a possible future fix but could easily be exploited as the personnel that are being used on the field would still be used in the audibles causing mismatches.

Any system implemented can possibly be exploited. The question is if this proposed system will have less of a negative impact on the "simness" of MFN than the current one that allows the same play to be called every down with minimal consequences.

Would it be exploitable or would it be an extra dimension to the cat-and-mouse game? Teams will frequently show one look to get the offense to behave a certain way only to change it up later in the game when their opponent makes an adjustment. Or they'll show one look for several weeks leading up to a key matchup just to throw that main opponent a curve ball in that game. Remember, you don't know what your opponent's game plan would be, so you might attempt to exploit what you think their audibles will be when in fact they call your bluff and scout a different play you've used successfully further back.

Does this mean the Shotgun runs had a positive average in your testing? I find that a little hard to believe, as I have only seen these plays get positive yardage a handful of times in over 3 years of playing MFN. Usually they are -3 yards in a cloud of dust.

Yes, they all had an average gain, the lowest was the HB Delay but it had good results against a bunch of plays (mostly deep zones)

What engine was this tested on? Bleeding edge or 0.4? I can't really comment if it's the new engine, but if not then I'm not remotely surprised every offensive play posted aggregate positives, because zone defenses are (currently) free money. I'm curious as to what they were successful in defending against, 'cause I've never faced anyone running lots of zone and not decimated them.

As to the proposed solution, I'm somewhat wary of the game 'telling you' what plays are best to run in a given situation.

You should also consider mirroring an existing team managed by a 'top' user for the purposes of gen'ing results, because how the sim determines overall ratings don't match the reality for many positions. For instance, when gen'ing the counter run plays, did the pulling guards have speed? Did DTs on any given play have speed? Were RBs at 95+ speed? Maybe you notice a recurring theme here, haha. I'm especially interested in this, because if all inside running plays you simmed were averaging 4.2 (roughly the average for rushing plays in the NFL) I'd be skeptical, 'cause inside running is practically non-existent at the higher levels of play.

Also, how many sims did each play get versus each other play? When we say 'all had a positive average', what type of range are we talking about? How sharp was the bell curve? Etc. After all, 0.0001 is technically positive, but if another play is averaging 10+, that's a problem.

Edit: Wanted to add one last question: Were these run with all three keys, or just neutral? Would be interested as well to see how keying impacted results, since there are several formations where there simply isn't sufficient play diversity to 'take advantage' of someone keying (think 1RB/4WR or 2RB/3TE for the most extreme examples of this).