BuzzerBeater Forums

Well, first of all thank you very much to all the people who took their time to answer my previous request.

I think there are some interesting conclusions to be taken from this analysis.

First, the potential balls do seem to be random, but they also seem to obey a well defined distribution. I won't enter into detail here (mainly because I wouldn't be able to :) ), but if all my assumptions are correct here's the chance you have to scout a player with a particular potential:

1 - 22.5%
2 - 40%
3 - 22.5%
4 - 10%
5 - 5%

By the way, if this is correct, the average potential is 2.35. But I think the distribution is more important.


Anyway, I think a more important and unexpected detail has emerged. There seems to be a correlation between the star rating and the potential of each player. I have very few statistics on this, but there definitely seems to be a tendency.
Look at the averages below, first number is star rating and the second is average potential for that rating:

1 - 1.6
2 - 2
3 - 2
4 - 2.5
5 - 3.2

Again, don't take these figures for granted. But I think you all see the pattern.

As a final disclaimer, please don't base your draft strategy on these numbers! The statistical pool is too small, and age hasn't been taken into account.
I just think it's a very good starting point, a teaser if you want, for anyone willing to start a more extensive data collection.

Wouldn't this be easier to do by looking at the results of past drafts? We have at least approximate boundaries for each ball and grade rating as it corresponds to salary, and a pretty good idea of what each potential is in ball ratings as well. You could look at a couple random leagues, and check out their draft history from last year to get as much data as you wanted.

Wouldn't this be easier to do by looking at the results of past drafts? We have at least approximate boundaries for each ball and grade rating as it corresponds to salary, and a pretty good idea of what each potential is in ball ratings as well. You could look at a couple random leagues, and check out their draft history from last year to get as much data as you wanted.

i don't think that the grade to salary rating is right, and why not taking the easy one ;)

For the study, i find total number for this conclusion fine?

Great work and I am seeing the correlation between star rating and potential as well.However, I only looked at 5 star and 4 star rated players. I got an average potential of 3.25 for 5 star rated players and a average of 2.82 for 4 star rated players. So it may be well worth taking the risk on players that have been scouted a second time.

i ask it a second time, how i achieved that work or how big is the sample?

As for the potential distribution, it was evaluated on a pool of little more than 100 players.
I feel quite confident with that however, because even after taking into account 50 or so players, the shape of the distribution was already well defined; the rest of the players only confirmed that.
Even without a methodical statistical test, I don't expect the real distribution to be significantly different from the one I posted.

The potential vs. star rating average, however, is little more than a guess. Not only is it more elaborate than a simple distribution, it is also based on a pool of 44 players.
It seems to be accurate enough, however. There's one guy, for example, who posted his result based on 14 players scouted twice. Here are the numbers he gave:

1 - 1 (2 players)
2 - 2.25 (4)
3 - 2.5 (4)
4 - 2.5 (2)
5 - 3 (2)

As you can see, it's not that far from the final results, and it does show a tendency even with so few players.

Actually, it would also be very interesting to evaluate the distribution of star ratings as well, because that would allow us to cross-check the results of all three calculations.
If you know the distribution of star ratings, the distribution of potentials, and the average potential vs. rating, you have three correlated elements that you can check for consistency.

As for the potential distribution, it was evaluated on a pool of little more than 100 players.

Even with 100 players, there must have been very few with 5 ball potential, no? Well, I am guessing 5 since you said 5%. Also, 10 4 ball players since you said 10%.
;-)

These are pretty small numbers - I would be hesitant to make any conclusions on how they match up to star rating.


Last edited by HeadPaperPusher at 12/10/2009 3:00:56 PM

Even with 100 players, there must have been very few with 5 ball potential, no? Well, I am guessing 5 since you said 5%.

Of course. But at the same time, there were more than 40 2-ball players. :)

Furthermore, there was only one lucky guy who scouted two 5-ball players, nobody else scouted more than one, even out of 14 or 15 players.

Of course, when I say that we have a 40% chance of scouting a 2-ball potential player, I can safely assume that that number can range from 35% to 45%.
5%, however, is a very low figure and therefore can vary wildly. Nevertheless, I can tell you that out of 10 analyzed drafts, in 8 cases the rate of 5-ball players was well below 10%. The only real exception to this rule is the one I already mentioned, with two 5-ball players scouted out of 10.

Additionally, the % I reported seem to fit a smooth distribution. Judging from the slope of the right tail of said distribution, I would say that the real 5-ball % may even be lower...

Additionally, the % I reported seem to fit a smooth distribution. Judging from the slope of the right tail of said distribution, I would say that the real 5-ball % may even be lower...

I think maybe I'm just not understanding the data well and not bothering to actually read it.