Skip to main content

A tool to automate and optimize DraftKings and FanDuel lineup construction.

Project description

Introduction · Build Status ·

An incredibly powerful tool that automates and optimizes lineup building, allowing you to enter thousands of lineups in any DraftKings or FanDuel contest in the time it takes you to grab a coffee.

Installation

Requires Python 3.9+.

pip install draftfast

Usage

Example usage (you can experiment with these examples in repl.it):

from draftfast import rules
from draftfast.optimize import run
from draftfast.orm import Player
from draftfast.csv_parse import salary_download

# Create players for a classic DraftKings game
player_pool = [
    Player(name='A1', cost=5500, proj=55, pos='PG'),
    Player(name='A2', cost=5500, proj=55, pos='PG'),
    Player(name='A3', cost=5500, proj=55, pos='SG'),
    Player(name='A4', cost=5500, proj=55, pos='SG'),
    Player(name='A5', cost=5500, proj=55, pos='SF'),
    Player(name='A6', cost=5500, proj=55, pos='SF'),
    Player(name='A7', cost=5500, proj=55, pos='PF'),
    Player(name='A8', cost=5500, proj=55, pos='PF'),
    Player(name='A9', cost=5500, proj=55, pos='C'),
    Player(name='A10', cost=5500, proj=55, pos='C'),
]

roster = run(
    rule_set=rules.DK_NBA_RULE_SET,
    player_pool=player_pool,
    verbose=True,
)

# Or, alternatively, generate players from a CSV
players = salary_download.generate_players_from_csvs(
  salary_file_location='./salaries.csv',
  game=rules.DRAFT_KINGS,
)

roster = run(
  rule_set=rules.DK_NBA_RULE_SET,
  player_pool=players,
  verbose=True,
)

You can see more examples in the examples directory.

Game Rules

Optimizing for a particular game is as easy as setting the RuleSet (see the example above). Game rules in the library are in the table below:

League Site Reference
NFL DraftKings DK_NFL_RULE_SET
NFL FanDuel FD_NFL_RULE_SET
NBA DraftKings DK_NBA_RULE_SET
NBA FanDuel FD_NBA_RULE_SET
MLB DraftKings DK_MLB_RULE_SET
MLB FanDuel FD_MLB_RULE_SET
WNBA DraftKings DK_WNBA_RULE_SET
WNBA FanDuel FD_WNBA_RULE_SET
PGA FanDuel FD_PGA_RULE_SET
PGA DraftKings DK_PGA_RULE_SET
PGA_CAPTAIN DraftKings DK_PGA_CAPTAIN_RULE_SET
NASCAR FanDuel FD_NASCAR_RULE_SET
NASCAR DraftKings DK_NASCAR_RULE_SET
SOCCER DraftKings DK_SOCCER_RULE_SET
EuroLeague DraftKings DK_EURO_LEAGUE_RULE_SET
NHL DraftKings DK_NHL_RULE_SET
NBA Pickem DraftKings DK_NBA_PICKEM_RULE_SET
NFL Showdown DraftKings DK_NFL_SHOWDOWN_RULE_SET
NBA Showdown DraftKings DK_NBA_SHOWDOWN_RULE_SET
MLB Showdown DraftKings DK_MLB_SHOWDOWN_RULE_SET
XFL DraftKings DK_XFL_CLASSIC_RULE_SET
Tennis DraftKings DK_TEN_CLASSIC_RULE_SET
CS:GO DraftKings DK_CSGO_SHOWDOWN
F1 DraftKings DK_F1_SHOWDOWN
NFL MVP FanDuel FD_NFL_MVP_RULE_SET
MLB MVP FanDuel FD_MLB_MVP_RULE_SET
NBA MVP FanDuel FD_NBA_MVP_RULE_SET

Note that you can also tune draftfast for any game of your choice even if it's not implemented in the library (PRs welcome!). Using the RuleSet class, you can generate your own game rules that specific number of players, salary, etc. Example:

from draftfast import rules

golf_rules = rules.RuleSet(
    site=rules.DRAFT_KINGS,
    league='PGA',
    roster_size='6',
    position_limits=[['G', 6, 6]],
    salary_max=50_000,
)

Settings

Usage example:

class Showdown(Roster):
    POSITION_ORDER = {
        'M': 0,
        'F': 1,
        'D': 2,
        'GK': 3,
    }


showdown_limits = [
    ['M', 0, 6],
    ['F', 0, 6],
    ['D', 0, 6],
    ['GK', 0, 6],
]

soccer_rules = rules.RuleSet(
    site=rules.DRAFT_KINGS,
    league='SOCCER_SHOWDOWN',
    roster_size=6,
    position_limits=showdown_limits,
    salary_max=50_000,
    general_position_limits=[],
)
player_pool = salary_download.generate_players_from_csvs(
    salary_file_location=salary_file,
    game=rules.DRAFT_KINGS,
)
roster = run(
    rule_set=soccer_rules,
    player_pool=player_pool,
    verbose=True,
    roster_gen=Showdown,
)

PlayerPoolSettings

  • min_proj
  • max_proj
  • min_salary
  • max_salary
  • min_avg
  • max_avg

OptimizerSettings

  • stacks - A list of Stack objects. Example:
roster = run(
    rule_set=rules.DK_NHL_RULE_SET,
    player_pool=player_pool,
    verbose=True,
    optimizer_settings=OptimizerSettings(
        stacks=[
            Stack(team='PHI', count=3),
            Stack(team='FLA', count=3),
            Stack(team='NSH', count=2),
        ]
    ),
)

Stack can also be tuned to support different combinations of positions. For NFL, to only specify a QB-WRs based stack of five:

Stack(
    team='NE',
    count=5,
    stack_lock_pos=['QB'],
    stack_eligible_pos=['WR'],
)
  • custom_rules - Define rules that set if / then conditions for lineups.

For example, if two WRs from the same team are in a naturally optimized lineup, then the QB must also be in the lineup. You can find some good examples of rules in draftfast/test/test_custom_rules.py.

from draftfast.optimize import run
from draftfast.settings import OptimizerSettings, CustomRule

# If two WRs on one team, play the QB from same team
settings = OptimizerSettings(
    custom_rules=[
        CustomRule(
            group_a=lambda p: p.pos == 'WR' and p.team == 'Patriots',
            group_b=lambda p: p.pos == 'QB' and p.team == 'Patriots',
            comparison=lambda sum, a, b: sum(a) + 1 <= sum(b)
        )
    ]
)
roster = run(
    rule_set=rules.DK_NFL_RULE_SET,
    player_pool=nfl_pool,
    verbose=True,
    optimizer_settings=settings,
)

Another common use case is given one player is in a lineup, always play another player:

from draftfast.optimize import run
from draftfast.settings import OptimizerSettings, CustomRule

# If Player A, always play Player B and vice versa
settings = OptimizerSettings(
    custom_rules=[
        CustomRule(
            group_a=lambda p: p.name == 'Tom Brady',
            group_b=lambda p: p.name == 'Rob Gronkowski',
            comparison=lambda sum, a, b: sum(a) == sum(b)
        )
    ]
)
roster = run(
    rule_set=rules.DK_NFL_RULE_SET,
    player_pool=nfl_pool,
    verbose=True,
    optimizer_settings=settings,
)

Custom rules also don't have to make a comparison between two groups. You can say "never play these two players in the same lineup" by using the CustomRule#comparison property.

# Never play these two players together
settings = OptimizerSettings(
    custom_rules=[
        CustomRule(
            group_a=lambda p: p,
            group_b=lambda p: p.name == 'Devon Booker' or p.name == 'Chris Paul',
            comparison=lambda sum, a, b: sum(b) <= 1
        )
    ]
)
roster = run(
    rule_set=rules.DK_NBA_RULE_SET,
    player_pool=nba_pool,
    verbose=True,
    optimizer_settings=settings,
)

Importantly, as of this writing, passing closures into CustomRules does not work (ex. lambda p: p.team == team), so dynamically generating rules is not possible. PRs welcome for a fix here, I believe this is a limitation of ortools.

LineupConstraints

  • locked - list of players to lock
  • banned - list of players to ban
  • groups - list of player groups constraints. See below
roster = run(
    rule_set=rules.DK_NFL_RULE_SET,
    player_pool=player_pool,
    verbose=True,
    constraints=LineupConstraints(
        locked=['Rob Gronkowski'],
        banned=['Mark Ingram', 'Doug Martin'],
        groups=[
            [('Todd Gurley', 'Melvin Gordon', 'Christian McCaffrey'), (2, 3)],
            [('Chris Carson', 'Mike Davis'), 1],
        ]
    )
)
  • no_offense_against_defense - Do not allow offensive players to be matched up against defensive players in the optimized lineup. Currently only implemented for soccer, NHL, and NFL -- PRs welcome!

CSV Upload

from draftfast.csv_parse import uploaders

uploader = uploaders.DraftKingsNBAUploader(
    pid_file='./pid_file.csv',
)
uploader.write_rosters(rosters)

Support and Consulting

DFS optimization is only one part of a sustainable strategy. Long-term DFS winners have the best:

  • Player projections
  • Bankroll management
  • Diversification in contests played
  • Diversification across lineups (see draftfast.exposure)
  • Research process
  • 1 hour before gametime lineup changes
  • ...and so much more

DraftFast provides support and consulting services that can help with all of these. Let's get in touch today.

Contributing

Run tests or set of tests:

# All tests
nosetests

# Single file
nosetests draftfast/test/test_soccer.py

# Single test
nosetests draftfast/test/test_soccer.py:test_soccer_dk_no_opp_d

Run linting

flake8 draftfast

Credits

Special thanks to swanson, who authored this repo, which was the inspiration for this one.

Current project maintainers:

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

draftfast-3.7.6.tar.gz (42.8 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page