AFL Fantasy Optimization
Budget: $250 – $750 AUD
Operations Research/Multiple Integer Programming Task:
My project revolves around optimizing an AFL fantasy football team, using a player's 2017 scores, and adhering to certain constraints. This allows me to verify the code delivers the correct solution.
My Objective Function: Maximize my team's season score, which is the score from Week/Round 1 all the way through to Week 23.
Bound to the following Constraints:
1. Pick 30 players out of a possible 652 players.
2. Pick players from different positions, such that there are 8 DEFs, 10 MIDs, 4 RUCs, and 8 FWDs. Some players can play more than one position.
3. Understand that only 22 out of the 30 players are on the 'field' - meaning their scores count towards the weekly team score. The remaining 8 players are on the 'bench', who's score doesn't count. The field positions are: 6 DEFs, 8 MIDs, 2 RUCs, 6 FWDs. The bench positions are: 2 DEFs, 2 MIDs, 2 RUCs, 2 FWDs.
4. Each player has a price. The total cost of all 30 players must be less than or equal to the starting budget of $12.6mil.
5. Each week, I only get 2/3 trades (2 most weeks, 3 over designated rounds). Therefore, out of the 30 players in the current week's team, at minimum 27/28 players should be the exact same as the previous week. Furthermore, each trade should adhere to my remaining salary cap budget from week to week.
Additional info: https://fantasy.afl.com.au/classic/help/game-guidelines (Rules may have slightly changed since 2017)
Input CSV File:
Columns: Player ID, Player Name, Week 1 Position, Week 1 Price, Week 1 Score, ... , Week 23 Position, Week 23 Price, Week 23 Score
All 652 rows, not including the header, will be filled with each player's projections.
Output CSV File:
Columns: Week 1, Week 2, ... , Week 23
All 30 rows, not including the header, will be filled with the Player ID of each week's selection.
Python Code:
I would heavily appreciate, as I am very code illiterate, if you could provide additional information that allows me to slightly alter the parameters and adapt the code for future season adjustments in the rules.
Last, but not least, as I will be running this optimization code not only before the season starts, but also during the season, it either needs to be able to factor in my past selection decisions.
Thanks for taking the time to read this.
My project revolves around optimizing an AFL fantasy football team, using a player's 2017 scores, and adhering to certain constraints. This allows me to verify the code delivers the correct solution.
My Objective Function: Maximize my team's season score, which is the score from Week/Round 1 all the way through to Week 23.
Bound to the following Constraints:
1. Pick 30 players out of a possible 652 players.
2. Pick players from different positions, such that there are 8 DEFs, 10 MIDs, 4 RUCs, and 8 FWDs. Some players can play more than one position.
3. Understand that only 22 out of the 30 players are on the 'field' - meaning their scores count towards the weekly team score. The remaining 8 players are on the 'bench', who's score doesn't count. The field positions are: 6 DEFs, 8 MIDs, 2 RUCs, 6 FWDs. The bench positions are: 2 DEFs, 2 MIDs, 2 RUCs, 2 FWDs.
4. Each player has a price. The total cost of all 30 players must be less than or equal to the starting budget of $12.6mil.
5. Each week, I only get 2/3 trades (2 most weeks, 3 over designated rounds). Therefore, out of the 30 players in the current week's team, at minimum 27/28 players should be the exact same as the previous week. Furthermore, each trade should adhere to my remaining salary cap budget from week to week.
Additional info: https://fantasy.afl.com.au/classic/help/game-guidelines (Rules may have slightly changed since 2017)
Input CSV File:
Columns: Player ID, Player Name, Week 1 Position, Week 1 Price, Week 1 Score, ... , Week 23 Position, Week 23 Price, Week 23 Score
All 652 rows, not including the header, will be filled with each player's projections.
Output CSV File:
Columns: Week 1, Week 2, ... , Week 23
All 30 rows, not including the header, will be filled with the Player ID of each week's selection.
Python Code:
I would heavily appreciate, as I am very code illiterate, if you could provide additional information that allows me to slightly alter the parameters and adapt the code for future season adjustments in the rules.
Last, but not least, as I will be running this optimization code not only before the season starts, but also during the season, it either needs to be able to factor in my past selection decisions.
Thanks for taking the time to read this.