dev diary #44 Overall And Attribute Grades

[Rocketcan]

Hello Coaches!

 

Welcome to another Dev Diary for NewSim on SimFBA!

 

Last time we talked about how Potential and Progression worked in the sim. This time, we are going to discuss how Overall and Attribute grades work in the college sim. You know, those pesky letters!

 

Overall Grade

 

The Overall letter grade of a player is calculated from a weighted average of Attributes that player has, depending on their position. You can tell which Attributes matter for a particular position based on which ones are listed for that player's position when you look at the player's card on the interface (except for stamina and potential, those do NOT go towards Overall calculations). The letter grade hides an integer value (between 0-99) underneath. The scale for Overall grades are as follows:

• 45+: A
  
• 35-44: B
  
• 25-34: C
  
• 15-24: D
  
• Below 15: F
  

Attribute Grades

 

Each attribute has a letter grade that hides an integer value (between 0-99) underneath. Unlike Overall, however, the attributes for a player are no longer graded on a flat scale (like anything above a 45 is an A, for example). We recently switched up the system to now grade a player against other players at their position across the entire NCAA. So, for example, a QB's Throw Power will be compared against all other QB's Throw Power in the NCAA.

 

Why did we make this change? A couple of reasons. First we realized that some skills like Speed were always an A for players at skill positions, and usually D or F for players at non-skill positions. While this reflected the reality of football that most non-skill players would lose in a race against a skill player, it did not actually provide any new information to the user. This leads us to the second reason. Within a position group like WR or RBs, there was no way to tell that a one player was faster than another, which is quite critical information. Thus, we needed a way to distinguish between players *within* position groups. Finally, not every Attribute is distributed among the same values. Thus, we would need to have different scales for each attribute anyway, so we needed to find one that takes care of the previous two problems.

 

To do that, we decided to calculate the average and standard deviation of each Attribute grade for each position. Then we use the following to determine which letter grade to assign for that Attribute (Number in parentheses is the rough percentage of players with that grade)

• Attribute is more than 2 standard deviations above the mean: A (2.28%)
  
• Attribute is more than 1 standard deviations above the mean: B (13.59%)
  
• Attribute is above the mean: C (34.13%)
  
• Attribute is within 1 standard deviation below the mean: D (34.13%)
  
• Attribute is more than 1 standard deviation below the mean: F (15.87%)
  

As you can see, having an "A" in an Attribute is VERY good. "B" is also quite good, as they are better than almost 85% of players at their position. Having a "C" is not bad at all, and actually signifies "average to above average", being better than at least half of all players at their position in that Attribute. D is not good, as it is below average, but not terrible. Remember, a player can be better than 49% of all other players at his position and still get a "D"! So don't fret if your freshmen have few Ds. That's probably fine. Finally, a definitely not good rating is F, at least compared to their peers at that position.

 

Given these statistics, most players will have a C or D rating at several Attributes, and that isn't necessarily bad. What this *does* do is signify who the truly good players are at a position, which is what we want. Sports are the result of a Pareto distribution: There are exponentially fewer good players than average ones.

 

One of the few things this system is not necessarily good at is determining whether a player is ripe for a position change or not. Because all of the Attributes are relative, just because a WR has a Throw Power of A, doesn't mean he'll have very good Throw Power once moved to QB (in fact he could very well have D or so Throw Power). While this is true, it does still give you *some* idea of whether or not a player might be good at another position (like a FB with A catching might be a decent TE). Position changes are inherently VERY risky, and this system models that well, I think.

 

One last thing: Stamina. This attribute is strange in that the distribution for it is a bit wider than most. It also is one that need to be applied to the sim differently than all others. So expect there to be a bit of a rework on Stamina in the future. That will most likely (as long as nothing blows up *figuratively*) make it into the NewSim engine changes for 2022.

 

Why no + or -?

 

In high school and college, the user can only see a letter grade without + or -. This is for a couple of reasons. First, on a team of nearly 100 players, it is pretty difficult to tell EXACTLY which players are better than others. Many players, especially at this level are fairly close to one another in talent, and the different between a 46 and a 47 is probably not noticeable when this player has been on everyone's radar for a few years.

 

Second, this is a big picture decision. The NFL is around the corner, and that is where the numbers hidden underneath the letters (except for potential) will be revealed. However, that will not occur until after the draft, and probably not even for a couple of weeks into the season. However, one of the things about the draft process (which is still WIP) will be scouting players to get a + or - attached to all of their grades (Attributes AND Overall). Thus giving NFL teams more information on a player in order to make better drafting decisions. I won't go any further into that, though, as that is a topic for another day!

 

That's all there is for this Dev Diary. Please ask questions below, and I will get to them as soon as I can.

 

Until next time,

 

Rocketcan

[Sarge]

Thanks as always for taking time to elaborate for us neanderthals.

[Rocketcan]

Live view from Rocketcan's office:

 

[tsweezy]

To be clear: when comparing attributes, they are only compared to the exact same position? So an OG with an A run blocking is only compare to other OGs NOT OTs, is that correct?

[Rocketcan]

Yes.

 

However, OGs, OTs, and Cs are probably going to be pretty close in terms of the breakdown of their skills.