Skip to content

Genetic algorithm library to generate daily fantasy sports lineups

License

Notifications You must be signed in to change notification settings

KengoA/fantasy-ga

Repository files navigation

fantasy-ga

Downloads Downloads

fantasy-ga is a Python module and a command line tool that uses the genetic algorithm to automate the generation of fantasy sports lineups. Currently supported platforms and leagues are as follows.

NBA NFL MLB NHL
DraftKings

Installation

Dependencies: numpy

pip install fantasy-ga

Usage

Python

LineupGenerator class supports csv files exported from daily fantasy sports platforms for a given contest.

from fantasy_ga import LineupGenerator
from fantasy_ga.configs import Site, League, ModelConfig, ContestConfig

data_path = "examples/DraftKings/NBA/DKSalaries.csv"
export_path = "examples/DraftKings/NBA/export.csv"

cc = ContestConfig(Site.DK, League.NBA)
mc = ModelConfig(
    # initial population of random lineups
    n_pop = 1000
    # number of evolutions to itererate breeding and mutation for
    n_gen = 16
    # number of children lineups to choose from two best lineups
    n_breed = 30
    # number of random mutations for each evolution
    n_mutate = 30
    # number of compound evolutions with additional random lineups
    n_compound = 5
)

model = LineupGenerator(cc, mc)
model.read_csv(data_path)
model.fit()
# If top_n is not specified, it will save max(500, number of total lineups) lineups sorted by scores
model.export_csv(export_path, top_n=3)

lineups, scores = model.get_top_n_lineups(1)
print(
    f"""
    [Best Lineup]
    Players: {[model.id_to_name[id] for id in lineups[0]]} 
    Salary Total: {sum([model.id_to_salary[id] for id in lineups[0]])}
    Expected FPTS: {scores[0]}
    """
)

Using custom numpy.array for player data

Alternatively, you can provide a numpy.array where the first 3 columns correspond to player ID, salary, fantasy points (FPTS), followed by position information. For instance, the columns correspond to id,salary,fpts,PG,SG,SF,PF,C,G,F,UTIL for DraftKings Fantasy Basketball.

If you would like to use custom numpy array for data matrix instead of csv files, you can do so by using the set_matrix() method. For example

cc = ContestConfig(Site.DK, League.NBA)
mc = ModelConfig() # use default
m = np.array(
    [
        [0, 6600, 36.46503, 0, 0, 0, 1, 1, 0, 1, 1],
        [1, 4200, 26.760368, 0, 0, 1, 1, 0, 0, 1, 1],
        [2, 3000, 4.38538, 1, 1, 0, 0, 0, 1, 0, 1],
        [3, 5000, 27.175564, 0, 0, 0, 0, 1, 0, 0, 1],
        [4, 3400, 16.734577, 0, 1, 1, 0, 0, 1, 1, 1],
        [5, 5900, 3.4382372, 0, 1, 1, 0, 0, 1, 1, 1],
        [6, 3000, -0.18490964, 1, 1, 0, 0, 0, 1, 0, 1],
        [7, 3000, 11.075589, 0, 0, 1, 1, 0, 0, 1, 1],
        [8, 3000, 6.469783, 0, 0, 0, 0, 1, 0, 0, 1],
        [9, 3000, 8.459954, 0, 0, 0, 0, 1, 0, 0, 1],
        [10, 5700, 33.98281, 0, 0, 0, 1, 1, 0, 1, 1],
    ]
)
model = LineupGenerator(cc, mc)
model.set_matrix(m)
model.fit()

Running individual steps of the Genetic Algorithm

LineupGenerator class has two core methods which return optimised lineups. breed() method chooses the best two lineups according to the sum of fantasy points with valid positions and swap players randomly (creating children lineups). mutate() method randomly swaps players where applicable. fit() method wraps around those methods such that those operations are done for multiple generations with additional random lineups.

CLI

As a Python module

$ python -m fantasy_ga --data_path=examples/DraftKings/NBA/DKSalaries.csv --export_path=examples/DraftKings/NBA/lineups.csv --site=DraftKings --league=NBA --n_pop=100 --n_gen=5 --n_breed=100 --n_mutate=100 --n_compound=10 --top_n_lineups=3

or a CLI command

$ fantasy-ga --data_path=examples/DraftKings/NBA/DKSalaries.csv --export_path=examples/DraftKings/NBA/lineups.csv --site=DraftKings --league=NBA --n_pop=100 --n_gen=5 --n_breed=100 --n_mutate=100 --n_compound=10 --top_n_lineups=3

which generates

Saved top 3 lineups into "examples/DraftKings/NBA/lineups.csv".

[Best Lineup]
Players: ['Reggie Jackson', 'Max Strus', 'Anthony Edwards', "Royce O'Neale", 'Nikola Jokic', 'Dejounte Murray', 'John Collins', 'Jarrett Allen']
Salary Total: 50000
Expected FPTS: 268.13