Optimizer for Daily Fantasy Sports • dfsOptimizer

Daily Fantasy Lineup Optimization in R

The dfsOptimizer package is intended to provide a straightforward way to generate optimal daily fantasy lineups for various sports, sites, and contests.

dfsOptimizer was built and tested in R 4.0.2

NOTE WELL: This is currently very much in Beta – I’m almost certainly not yet exporting everything I need to, and new issues always crop up once someone who isn’t actually developing the thing gets their hands on it. So, if you decide to use it, I’d very much appreciate any bug reports and/or suggestions!

Installation

devtools::install_github("anthonyshook/dfsOptimizer")

To-Do

Unit Testing
More sports and sites
Add CUSTOM class and API so users can add their own sports

Included Sites/Sports

Sport	Draftkings	FanDuel	Yahoo	Draftkings Single Game	FanDuel Single Game
HOCKEY (NHL)	+	+	+	+	-
FOOTBALL (NFL)	+	+	-	+	-
BASEBALL (MLB)	+	-	-	+	-
BASKETBALL (NBA)	+	+	+	+	-
GOLF	+	+	+	-	-
NASCAR	+	+	-	-	-
WNBA	+	-	-	-	-
SOCCER	+	-	-	-	-

Notes

The package relies on the ROI package, and comes with one solver as a dependency – ROI.plugin.glpk. This is to ensure it works out of the box, so to speak. However, the solver argument in build_lineups() is compatible with the ompr.roi package, so if you have ROI.plugin.symphony installed, it should be as simple as `build_lineups(solver=‘symphony’)

Simple Examples

Build Lineups from a Draftkings roster file (i.e., that you download from DK). This example uses the mean Points-per-game provided by draftkings as a projection, which I don’t recommend you actually do.

Quick Way - pass file in creation statement

# Build an optimizer
mod <- create_optimizer(site = 'DRAFTKINGS', 
                        sport = 'HOCKEY', 
                        contest_type = 'CLASSIC', 
                        filepath = '/Path/to/DKSalaries_nhl.csv')

# Build 20 lineups
lineups <- build_lineups(mod, num_lineups = 20)

# Get summary stats (player and team exposure breakdowns, and a cross-lineup similarity measure)
summary(lineups)

# See the lineups
lineups

# OR save them to a list
lineups_extracted <- as.list(lineups)

# OR export them into a format that you can upload to the selected site
export_lineups(lineups, file = '/path/to/lineup_export.csv')

Slightly less quick way - parse file and add players (you could use this to replace the player list on an already build model)

# Build an optimizer
mod <- create_optimizer(site = 'DRAFTKINGS', 
                        sport = 'HOCKEY', 
                        contest_type = 'CLASSIC')

# Parse and add players
mod <- add_players_from_csv(mod, filepath = '/Path/to/DKSalaries_nhl.csv')

You can also add players from a data.frame – see ?add_players_from_df or ?get_players_from_data_frame for more detail and examples!

Updating the Projected Fantasy Points for your Model

Once you have an existing optimizer model, you can update the FPTS using the function update_fpts. This requires a data.frame that contains columns id and fpts – the easiest way to get players’ ids is using get_player_data on an optimizer with players added.

# Build an optimizer
mod <- create_optimizer(site = 'DRAFTKINGS', 
                        sport = 'HOCKEY', 
                        contest_type = 'CLASSIC',
                        filepath = '/Path/to/DKSalaries_nhl.csv')

# Get the data
df <- get_player_data(mod)

# From here, you can replace the value of FPTS in df by either
# merging with another table by player name or ID
# As an example, we'll just toss in some random values
df$fpts <- runif(nrow(df), 0, 1)

# Update the model with the new FPTS
mod <- update_fpts(mod, fpts_data = df)

README