Two easy changes that can help CFM

[adembroski]

So, I'm suggesting these because I'm about 90% sure they can be done in CFM with little to no art or engineer assistance. I'm not 100% on the latter, but they certainly would not be major tasks and would be mostly designer asset adjustments and it's all under the hood.

Both would significantly improve the user experience in CFM and the first would improve CPU roster management.

1. Normalize attributes by position before calculating the weighted average for OVR
As it stands, the overall rating is calculated using a weighted average of select attributes, which is then normalized bring the final overall number within a select range.

The problem with this method is that not all attributes are created equal. An 80 speed is OK for a TE, but it's crap for a CB, but yet it remains a net positive for the weighted average created for the position.

By simply using a formula to create a position-specific percentile rank in each attribute before performing the weighted average calculation, positionally weak attributes will have the appropriately negative effect on the overall.

I believe this is the real problem that inspired the 90-speed floor for regression, but I believe that solution is treating the symptom rather than the problem.

2. Use position-attribute specific percentile ranks for draftee grades- as compared to ALL potential rookies
You will never see a QB with an "F" in THP, even if he has the weakest arm in all of the draft classes. Why? Because it only uses the raw rating to grade. Would you honestly consider a 75 THP a "C"? No, it's an F. Hell, when I'm playing, I consider 80 to be the very lowest THP I'll accept (and he better be DAMN accurate).

The grades should tell us how good/bad they are in that rating compared to other rookies at that position. It increases granularity, allowing more interesting and informed decisions.

[SyncereBlackout]

Love the second idea. The first I suspect will cause a riot (like all things ratings)


Sent from my iPhone using Tapatalk

[adembroski]

Quote: Originally Posted by SyncereBlackout

Love the second idea. The first I suspect will cause a riot (like all things ratings) Sent from my iPhone using Tapatalk

The visible attributes remain the same, the percentile rank is used under the hood to perform the overall calculation only.

Nothing really changes except OVR becomes a more accurate reflection of the player's ability compared with others at his position.

Sent from my SM-G930P using Operation Sports mobile app

[deu22ces]

Is the normalization why if you put all the ratings for a player that are supposed to count toward OVR down to Zero, the player is still a 12 or 14 OVR (don't remember which)?

[adembroski]

Quote: Originally Posted by deu22ces

Is the normalization why if you put all the ratings for a player that are supposed to count toward OVR down to Zero, the player is still a 12 or 14 OVR (don't remember which)?

Yes. It is just a way to convert the weighted average to conform to the range desired. It gets tweaked every Madden.

Sent from my SM-G930P using Operation Sports mobile app