JeffHCross Project #1: How is OVR determined?

[psuexv]

yeah I understand what you guys are saying, but I don't completely agree. If we were truly trying to "predict" which variables had an affect on OVR then yes those insignificant variables would be disregarded.

However I think those other variables are definitely part of the formula so should be included. My rationale is, if they are not in the formula... why are they there. I would understand having them there as a placeholder so to speak since those variables are used in for other positions and not specific to each position, but why give them random ratings. Why is CBs spin move 83 if it's not factored into his OVR.

The other thing that I noticed is that there are 2 variables that are not included in any of the formulas the Rudy did. Return and Stamina. Return makes sense but Stamina?

[Rudy]

But psu, does it make sense to include a variable in the formula that is completely contradictory to value? Like a better spin move resulting in a lower overall rating? That just makes no sense and is completely wrong in a regression analysis.

The point of the exercise wasn't to come up with a new formula and create a way to say how good a player is at each position. The point was to determine what variables EA is using to get that overall rating. It's up to you then to determine how much that overall means to you. For example, I believe guards are graded higher for run block than pass block. If you are a pass happy Air Raid team you might want to overlook the overall rating in this case. Very clearly EA uses specific variables for specific positions. To use all ratings for every position is simply wrong. It would be the same as using the number of the player or his name. They shouldn't be correlated and even if there was a slight correlation we wouldn't bother including it.

[psuexv]

Rudy, I completely agree with you, I'm not saying what you are doing it is wrong. If I'm trying to determine which factors make my crops grow better and I find that timing between fertilizing has an insignificant effect, I'm not factoring that into my equation. However I for some reason feel EA is using all variables though, even though it may not make sense... gut feeling.

I tried to find some players that have identical "significant" ratings. That was kinda difficult. I did find something though that may help my case. 2 Kickers with identical KAC and KPW ratings. 84 and 84. Their awareness is 50 and 80 respectively. Based on your formula they should come out to be 82 and 84, when actually their ratings are 83 and 84. Yes it's a small difference but I think that it does show that their are some other factors coming into place.

I ran them with all variables included and got 83 and 84. In the grand scheme of things does it really make any difference, probably not, but I just have a feeling that they are using all variables.

[psuexv]

Well I was wrong, my damn gut anyway. EA does it right

[Rudy]

By the way, that 197-198 pound run stopping DE was the #11 DE in the country and a 4* player. Just looked him up the other day.

[psuexv]

So here's how I proved myself wrong . I created a player with the exact same ratings as a current player on the default roster, then I ran all of the non significant attributes from Rudy's equations up to 99. The OVR stayed the same which would lead us to the fact that, yes only certain attributes are in the OVR equation.

[Rudy]

I think the reason the overalls are off on some guys (1 point max) is due to the way they round the calculations and/or some other ratings that might seem insignificant but could be.

[ram29jackson]

So here's how I proved myself wrong . I created a player with the exact same ratings as a current player on the default roster, then I ran all of the non significant attributes from Rudy's equations up to 99. The OVR stayed the same which would lead us to the fact that, yes only certain attributes are in the OVR equation.

i'm just saying that I noticed that years ago.. I dont care or know what the significance is, just saying I noticed it

[JeffHCross]

yeah I understand what you guys are saying, but I don't completely agree. If we were truly trying to "predict" which variables had an affect on OVR then yes those insignificant variables would be disregarded.

That is what I was trying to predict

I think the reason the overalls are off on some guys (1 point max) is due to the way they round the calculations and/or some other ratings that might seem insignificant but could be.

This is true. One thing I know they do is cut off the OVR rating at 99 (i.e. some players are actually higher than 99, but the game displays 99) and, I believe, 40. The other thing I know is done is rounding.

[nmccauley]

Interesting thread. Has anyone devised a method (aside from Jeff) to derive Trucking, Elusiveness, Spin Move attributes based on actual player characteristics (i.e. HT, WT, 40 time, etc.) or actual stats (yards/carry, TD/carry, etc.)?

[JeffHCross]

Interesting thread. Has anyone devised a method (aside from Jeff) to derive Trucking, Elusiveness, Spin Move attributes based on actual player characteristics (i.e. HT, WT, 40 time, etc.) or actual stats (yards/carry, TD/carry, etc.)?

As far as I'm aware, no. And it's tough to characterize my method as deriving it either. I derive some, but not much. I'm hoping to change that in the future, but haven't had time yet.

The only other people I know that are looking into this kind of stuff are Playmakers (on OperationSports) and MattUM2. I know Playmakers makes his historic teams by finding a contemporary comparison for a historic player. I'm not fully up to speed on how Matt comes up with his numbers.

[twookool]

maybe i missed it, but where do you get the numbers to plug into the calculations?

[JeffHCross]

maybe i missed it, but where do you get the numbers to plug into the calculations?

They come from a player's individual ratings. SPD, AGI, THP, etc. You could also use this to estimate a recruit's incoming OVR (which is part of what PocketScout has done), by first converting the B+ and C- ratings to numbers.

The real meat of this regression isn't to find out an individual's OVR -- that's already there by a simple glance at the roster. It's what goes into that calculation that was interesting.

If someone wanted to get really sophisticated, they could now take something like what Rudy's done and modify it to their own liking. Essentially re-calculation the OVR ratings for their players. Think Play Recognition is more important for a defensive player than Block Shedding? Easy change.

[supersigmaof21]

Good info here!!

[JeffHCross]

Thanks for the reminder, Sigma. I really need to get back into this again. There's a few things I wanted to do but never had a chance to put together.

[IBI]

Yeah, thanks sigma. This was the first time I've read this thread, and there is a lot of great info in here. Jeff, while reading this thread, I was thinking exactly what you're suggesting. By being able to download my OD rosters, I could easily change the formula to weigh the attributes that I find important for a user controlled OVR, and set my lineup based on my own calculations, rather than sifting through all of the attributes. It will also help greatly with identifying my type of players in recruiting.

Excellent thread topic! I may run a test when I get home tonight, but does anyone know if Rudy's calculation is accurate for NCAA12?

[JeffHCross]

does anyone know if Rudy's calculation is accurate for NCAA12?

I didn't really ask directly while I was at the Community Day events last season, but ratings did come up several times in conversations between me and a couple devs that we were working on rosters/ratings. What I got out of those conversations was that there was a slight change to how a WR's overall was being calculated, but not significant.

I can't make any such prediction about 13 though.

[supersigmaof21]

I definitely think my year 3 FAU team is the better for me stumbling across this old thread. I think AWR should be more important (I play coach mode) but I understand why it's not because the user is the AWR for the players. Then again, what about the CPU controlled opponent.


Sent from my iPhone using Tapatalk

[JeffHCross]

The user's AWR really only takes the place of the QB (or HB on a running play) and the defender you choose to control. AWR does factor in to other players, IMO.

I think AWR is as important as it should be. As it stands now, you can't have a super OVR if you have high AWR but lack the physical skills. But you can be at least decent OVR if you have the physical skills but lack AWR. That's true for real football too ... you'll usually play a gifted player, who may make some mistakes, over the guy that understands the Xs and Os but that can't keep up with those around him.

But for you, since your playing CM, you should probably prioritize AWR when you're comparing your QBs. Shouldn't make much difference for how you assess other players though.

[supersigmaof21]

That's the funny thing ... All of my QBs AWR is sub 70 but the kid with the highest THA performs the best of the 3. I just suffered through 2-3 games without my starting QB and went 2-1. In those 3 games my backs definitely shouldered the load to give us a chance.


Sent from my iPhone using Tapatalk