NFL Style Defense Playbooks for Madden 08

[solarkev]

Recently, I’ve been trying to modernize playbooks to use with Madden 08. I made the defense playbooks in this thread (https://www.footballidiot.com/forum/viewtopic.php?f=15&p=153299&sid=2bce77b83dca8a7cad404bb127149dd5#p153299) using Madden 25 plays and formations (from madden-guides), with proper play tags so that the AI uses them well.

While I think that the AI will call plays for appropriate situations, the so-called “style” of the playcalling may not match NFL usage schemes. So, I would also like to meld actual NFL football styles into the playbooks. For such a purpose, I found a method to define defense “style” for NFL teams. Take note of the research displayed in the 3 graphs below These were reproduced from data taken here: https://www.fantasypoints.com/nfl/articles/2024/nfl-coverage-bible-broad-coverages#/

“Style” can be defined by playcall percentages for Zone%, Safety configuration% and Blitz%.

My method to reproduce NFL “style” with Madden 25 type playbooks for Madden 08

I loaded my playbooks into the Madden database editor and exported the PBPL and PBAI tables into Excel. The PBPL table has the play names, play number designator, and formation numbers. The PBAI table has the play number designator, PBAI tag, and playcall percentages for each play in each PBAI tag. I went through each play in the PBPL table and added designations for three categories: (1) Zone vs man, (2), 2 Hi vs 1 Hi safety locations, and (3) Blitz or not. Using the play number designators, I was able to associate these designations with each play in the PBAI table. Then, I was set up to write Excel formulas to modify the play call percentages to achieve various “style” percentages, like in the above graphs. For example, I could make sure that the top blitz call percentages would be about 30%, or the medium zone% would be about 70% (comparing to the graphs). After making appropriate percentage changes, the resulting PBAL table can be imported back into the playbook.

I decided to divide the teams in each graph into parts, treating 3-4 and 4-3 based teams separately. The top quarter teams were designated “H” for high. The middle half was designated “M” for medium. The lowest quarter teams were designated “L” for low. For example, a 4-3 team in the upper quarter for zone defense %, the middle half for 2 high safety formation %, and lowest quarter for blitz % would be called “4-3 Z-H 2H-M, B-L”. In this example, the “4-3 Z-H 2H-M B-L” would be appropriate for the Colts or the Jags and represent teams that tend to use more zone coverage and less blitzes than the average NFL teams. I did this so that I could label playbooks with usage expectations of the AI relative to other teams, but correct for the NFL.

Attached in the post following are several playbooks made using the above methods. The 4-3 playbooks all have the same plays and formations, but the playcall percentages vary to emulate the “style” in the playbook names.

[solarkev]

Playbooks described in the first post

[solarkev]

Another playbook (I can only upload 3 per post, I guess)

[lorak]

@solarkev,
Interesting information.
For example, your 43D Z-M 2H-M B-M defensive playbook is suitable for the Dolphins or Falcons, and possibly for a few other teams too.

Could you please elaborate on the play calling percentages?
For example, assume we have four plays for a given AIGR, with the following values in the "prct" field:

Code: Select all

AIGR PBPL prct       AIGR PBPL prct
5    1    10         5    1    20
5    2    20         5    2    30
5    3    30         5    3    40
5    4    40         5    4    50


Edit: Fixed the field names.

Considering the data from the two tables above, what would be the actual CPU play calling percentage for each play?

[solarkev]

Hi lorak,

Answering your question. If these four plays were the only ones with the AIGR tag (5, in this case), then the playcalling probabilities would be (rounding to closest whole number):
PLYT 1 - 20/140=14%
PLYT 2 - 30/140=21%
PLYT 3 - 40/140=29%
PLYT 4 - 50/140=36%
total 100%

The field in the PBAI table is labeled "prct"; I assumed that your per was for this, meaning percentage. Also, the PBAI table numbers plays using the label "PBPL" instead of PLYT.

The computer totals the prct values for all plays with a certain AIGR and calculates percentages from that total. I found this out by looking at original Cover 2 playbooks with the database editor in the default Madden 08.

Make sense?

[lorak]

Hi lorak,

Answering your question. If these four plays were the only ones with the AIGR tag (5, in this case), then the playcalling probabilities would be (rounding to closest whole number):
PBPL 1 - 20/140=0.14
PBPL 2 - 30/140=0.21
PBPL 3 - 40/140=0.29
PBPL 4 - 50/140=0.36
total 100%

The field in the PBAI table is labeled "prct"; I assumed that your per was for this, meaning percentage. Also, the PBAI table numbers plays using the label "PBPL" instead of PLYT.

The computer totals the prct values for all plays with a certain AIGR and calculates percentages from that total. I found this out by looking at original Cover 2 playbooks with the database editor in the default Madden 08.

Make sense?

Thanks for the answer, it makes sense to me.
Thus, for the first table above, the play calling probabilities would be 0.10, 0.20, 0.30 and 0.40 for PBPL 1, 2, 3 and 4 respectively.

Yes, I am refering to the "prct" field and "PBPL" as id to each play in the PBAI table. I just typed the wrong field names.