Dividing Data into Equal Groups

[montyjk01]

Hello,

I am trying to build a program that pairs golfers equally in groups of 4 and 5 players based on their "power ranking". The query selects the players that are playing that day as follows:

[Player] [Power Ranking]
Name1 50
Name2 45
Name3 37
Name4 36
Name5 28
Name6 25
Name7 23
Name8 18
Name9 11



I would like to have the data grouped so each group has about the same amount of ranking points. In this example, there would be one group of 4 and one group of 5. Any idea how to do that?

[Micron]

How about letting us know what you see as being those 2 groups for a start.
Being a bit of a duffer myself, I'm surprised a course would allow a group of five.

[isladogs]

Excuse what may be a dumb question and nothing to do with golf, but how can you pair an odd number of people - in this case 5?

[Micron]

It would be a five-some and a four-some. Usually groups of five are not allowed. I know I don't like playing behind one. Rather, this should be 3 groups of 3 as opposed to groups of 4 and a single. This is probably for tournament play.
EDIT
Unless this is another one of those posts where the info and the reality are not one and the same.

[montyjk01]

It would be a five-some and a four-some. Usually groups of five are not allowed. I know I don't like playing behind one. Rather, this should be 3 groups of 3 as opposed to groups of 4 and a single. This is probably for tournament play.
EDIT
Unless this is another one of those posts where the info and the reality are not one and the same.

It's for the daily money game played by members. Each group consists of 4 or 5 players. Since money is at stake the teams need to be divided as equally as possible among those playing. Obviously, the handicaps range from 0-18+ hence the calculation of power rankings based on birdie potential, score, etc. I have formula's for calculating the "Power Ranking". However, I'd like access to automatically pair the players in groups so that the sum of their power ranking is as close as possible.

I have figured out how to rank each player and then I can build a report that selects a player based on the rank. However, that often times does not pair the players together in a way their sums are close together. I can't figure out how the computer can select the best grouping so that the game is as fair as possible. Does that make sense?

I'd be willing to share the database if needed.

[Micron]

I played with this for quite some time but when you didn't post back, I didn't save my spreadsheet. Also did quite a bit of research looking for math solutions but none were of much use. In the end, I figured there were ways to get reasonably close, and even ways to tweak it but I abandoned it all when you didn't post back. I also concluded that equal grouping on a numerical total doesn't guarantee fairness, especially if you can have unbalanced groups. I saw that it's possible to put more lower handicap players (if that's what your numbers represent) in one group compared to the other yet be within 3 points of each other group-wise. I also found that more than one very close balance could be achieved, but not without impacting fairness. To be honest, I think you'd be better off using a spreadsheet and shift the numbers around while displaying a total. I'd start with allocating along the lines of (I'm using the descending order here)

1 2
4 3
6 5
8 7
9
and then look at the difference in column totals, then find 2 numbers to shift around where those numbers differ by half of the difference in the totals.