Draft Experiment (updated three times- thrice?)

[JT_55]

Seeing as no other person in this forum has posted a draft experiment in Live 2005, I will do so myself. I noted down the letter grades, potential level, and other info on selected players (projected top 5 and all excellect potential players).Then, I simmed the Dynasty to the end of the season and then, after the draft, I noted down the selected players' rating for each category.



The letter grades of the players are as follows:





After Draft Day and still in the offseason, the overall ratings for each player were:
Askins: 60 (1st overall)
Terrel: 63
Heller: 57
Reirden: 53
Otter: 55
Royal: 51 (1st round, 23rd pick)
Garvey: 44 (2nd round, 53rd pick)

Just by looking at the overall ratings, you can tell that the players with better potential (Terrel and Otter) have a better rating than the player projected to be drafted one spot before them (Askins and Rierden, respectively).

Here are the ratings in each category:





That's all the raw data I have. I'll update it again for the statistics.

[asgsjb]

Interesting stuff, JT. I appreciate your punching this up there. It seems like ranges can probably be derived from your stats, although there may be an occasional hiccup (i.e. Terrel's 71 in steals is a B, but Askins' 71 in offensive rebounding is a B+...how can the same score be both a B and a B+, unless there can be a slight difference for different categories, like steals and offesnive rebounds).

I don't see any pure A's or any A+'s, but the A-'s (all in conditioning) range from 78-89.

B+'s range from 71-79.

B's range from 65-81 (interesting, as it crosses over the range of B+'s)

B-'s range from 61-74.

No C+'s.

C's range from 48-67.

C-'s range from 43-61.

No D+'s.

D's range from 37-57.

D-'s range from 20-49.

E's range from 7-28.

So definitely some crossover between these ranges. It does seem clear that you do not want less than a B (for the most part) in any key category for a position (i.e. passing for a PG, blocks for a C, etc.) and you want the highest potential possible.

Again, thanks for the good work, JT.

[JT_55]

No problem. I'm always glad to help out fellow NBA Live players.

As you've said, there are occasional times where the grade is completely off from the rating. So to find out what the expected rating for each grade level are, I have determine the standard deviation by punching in every single number into my graphing calcuator. Then the SD should be a good way to figure out the expected range of ratings (Average rating-SD to average rating+SD). If there are any math experts out there that knows a better way, can you please tell me how?

I'll update it again when I'm done.

Update:

The number of times each letter grade showed up in my experiment:

A+ : 0
A : 0
A- : 9
B+ : 4
B : 12
B- :16
C+ : 0
C : 25
C- : 19
D+ : 0
D : 21
D- : 21
E : 6

As asgsjb pointed out, there were no A+'s, A's, C+'s, nor D+'s in my experiment (although that doesn't necessarily mean that there will be none that shows up in the game).

The average will give you a preliminary guess at what the rating will be for each letter grade. I put a star next to some of the averages not as accurate as they have too little information. Here are the averages:

A- : 83.89
B+ : 73.25*
B : 72.25
B- : 66.81
C : 58.72
C- : 51.67
D : 46.05
D+ : 37
E : 17.33

I added and subtracted the standard deviation from the averages to get a more advanced range to predict where the rating will be. The standard deviation and the expected range are as follows:

A- : 3.66, 80 to 88
B+ : 3.44, 70 to 77
B : 5.54, 67 to 78
B- : 3.57, 63 to 70
C : 4.92, 54 to 64
C- : 4.63, 47 to 56
D : 5.03, 41 to 51
D- : 6.26, 31 to 43
E : 6.5, 11 to 24

A chart of the above information:



The standard deviation can also be used to calcuate how evenly spread out the values are (or not). Note that this formula is not mathematically proven. It's just a method I like to use to find the "spreadoutness" of the values.

First, I take find out the SD of the number of values in a given set (say, the A- set). This is really hard to explain so I will use an example. So we have the A- set of values. The SD for that is 3.66. There are 9 numbers in that set so I find out the SD for any 9 consecutive numbers (1, 2, 3, 4, 5, 6, 7, 8, 9) and it truns out to be 2.58. Since the SD for the A-'s is more than the SD for the 9 consecutive numbers, the values are farther from the average than any 9 given numbers with the same average.

So its like:
1, 2, 3, 4, 5, 6 has average of 3.5
SD of that is 1.707

1, 1, 1, 6, 6, 6 also has average of 3.5
But SD is 2.5

3, 3, 3, 4, 4, 4 again has average of 3.5
But its SD is 0.5

Sort of get what I mean now? Sorry if this is all flying over your head but its the best I can do.

[JT_55]

To make a long story short (or even longer, depends on the way you look at it), I wrote down the letter grades for seven rookies and compared it to their ratings, along with their potential. I found out that a grade of A- will likely result in a rating of 80-88, B+: 70-77, B: 67-78, C: 54-64, C-: 47-56, D: 41-51, D-: 31-43, E: 11-24.

The overall rating of the players were:
1st overall (limited potential): 60, gained 4 after off-season and lost 2 next season.
2nd overall (good potential): 63, gained 4 after off-season and gained another 3 next year.
3rd overall (average potential): 57, gained 4 after off-season and gained 2 next season.
4th overall (average potential): 53, gained 4 after off-season and gained 3 next year.
5th overall (good potential): 55, gained 4 after off-season and gained 3 next year.
23rd overall (excellent potential): 51, gained 4 after off-season and gained 3 next year.
53rd overall (excellent potential): 42, gained 2 after off-season and 3 next season.

Moral of this story is: Don't draft limited potential players! Or else you risk them losing ratings.

[asgsjb]

Great stuff, JT...thanks! I'll be printing this out to hold onto for when I get to my draft. I appreciate it.

[JT_55]

Updated to incorporate charts.

[aussiesta]

Very interesting stuff. That means the best player in that draft just reached a 70 average, which is not that much really. I'm playing a five-season dynasty and none of the 'fake' draftees has ever reached 70; 69 is the top ranking among them, even after all this time, and most number-one picks are at around 64.

[JT_55]

Thanks. The rating depends mostly on the draft depth, I think how most of your draftees are around 64 is just due to weak drafts. Rumor (from this forum) has it that all draftees don't go over 75 overall rating when drafted. As with the information I provided in this experiment, might not want to count on it as I did a different one (with weak draft depth), and the averages are all very different. Don't bother to post it cause no one cares anymore, including me (I haven't played Live for a while now).

[Remiokas]

Very nice chart, but the thing that Drafties can only reach 70 or something around that is not true. I had an Undrafted rank player go up to 91 or 92 not sure realy, so i tried drafting undrafted rank players but the max ovr i got in a excelent potentian player in that category (Draft depht very stron) was around 50, so i think the drafties can become very good if u spend a lot points on there training.