Create team & match lineup for sports event


I am trying to get chatGPT to create teams and matches for a sports event. It never gets it right. There is always a small error. If I ask it to fix this error, it introduces another error and so on.

Any suggestions on how to get ChatGPT to get this right?

Here is my current prompt:

You are an expert team coach. The task is to create teams and a number of matches using the following rules:
A macth has 2 teams with 3 players and 1 pro each.

Player Information Setup:

List all players with their attributes: name, player type, player strength, match speed, number of macthes requested, and any player requests.
Define Match Structure:

Define the structure of each match. A match consists of two teams (‘white’ and ‘blue’) of 3 players and a pro each. Ensure each team has a pro. Fast players cannot be mixed with slow players within a team. If a team has only slow players, it is a slow team; if a team has only fast players, it is a fast team.

Assign Pros to macthes:

Assign the pros to each match, ensuring each match has at least one pro per team.
Separate fast players from slow players.

Create Initial Team Assignments:

Calculate the total number of matches by summing up the “requested number of matches” column in the table. Verify the sum is correct. Divide by 8 and round up to get the number of matches needed.
Create initial team assignments for each match. Ensure the following:
No fast and slow players in the same team.
Fast teams cannot play against slow teams.
Players cannot switch teams and colors between matches.
Assign fillers as players if there are not enough players for a team based on the requested number of matches. Players cannot be fillers. Fillers cannot be pros.
If a player plays only in two matches, they have to pause for one match after their first.
Ensure Player Participation:

Ensure each player is assigned to the correct number of matches according to their requests.
Adjust for Personal Requests:

Adjust the team assignments to accommodate personal requests (e.g., players who want to play in the same match). Verify that the adjustments do not break any of the previous rules.
Balance Team Strength:

Ensure that the aggregate team strength is balanced across all matches. Adjust the assignments if necessary to maintain balance.
Final Verification:

Verify the final match assignments to ensure:
No fast players are mixed with slow players within a team.
Fast teams do not play against slow teams.
Each team has a pro.
Verify that each player is assigned the exact number of matches requested.
Personal requests are respected.
Team strengths are balanced.
Adjust if you find errors.
Produce Final Match Breakdown:
Expand the original table and fill the “Assigned Matches” column G in the table with the calculated match numbers per player and fill the “team Color” column H in the table with the calculated team color.
Produce the final Match Breakdown, listing the teams for each match with player names, strengths, and speeds. Ensure all conditions are met and provide the final assignment table with match numbers and team colors for each player.

Player signup input table:

John player 7 fast 4 play all requested chukkers with Jared
Maggie player 3 slow 5
Steff player 6 fast 2
Mike player 8 fast 3
Sonja player 3 slow 4 play late chukker
Alex player 3 slow 4
Charles player 4 slow 4
Sandra player 2 slow 2
Caroline player 8 fast 4 play all requested chukkers with Santos
Jessica player 2 slow 4
Jules player 6 fast 2
Susi player 3 slow 4
Anton player 6 fast 2
Tom player 3 slow 2
Peter player 4 slow 2 play all requested chukkers with Christian
Olivia player 9 fast 2
Steve player 6 fast 2
Philipp player 6 fast 2
Adam pro
Pablo pro
Nic filler
Carlos filler
Mike filler

Welcome to the community!

GPT-4 is pretty amazing. Actually, I mean a lot of things are amazing - the human brain is a gigantic marvel of the universe. While both are suited to a wide variety of tasks (and I’m sure that your task could be accomplished as requested with the right prompt) - an LLM probably isn’t the best option for tabulating things.

Your prompt (I’ll admit I didn’t read the whole thing) seems to be quite algorithmic - why not instead use ChatGPT to help you write a reliable program that will accomplish the task for you?

you can use this chat as a starting point: ChatGPT

note however, that it’s wrong. Know that ChatGPT can only process a limited number of concepts at once - try to structure your instructions in a more sequential, less convoluted way (almost pseudocode) if it fails until you have your correct program.

It may take a bit of practice but I’m sure you can do it :slight_smile: