FPLOptimiseR

Installation

You can install FPLOptimiseR from github with:

# install.packages("devtools")
devtools::install_github('Chrisjb/fploptimiser')

Usage

fetch data

Get the historic player data from the FPL API for your own analysis

library(FPLOptimiseR)

df <- fetch_player_data()

Get expected points (adjusting for xG, xA and xCS) data from understat

df_xp <- fetch_xg_data()

Get expected clean sheets calculated with understat data:

Xclean_sheets <- fetch_xCS()

team	xCS_per_game_h	xCS_per_game_a	total_xCS	value
Leicester	0.44 (8)	0.34 (8)	6.2	7 (0.8)
Liverpool	0.4 (8)	0.37 (8)	6.2	3 (-3.2)
Manchester City	0.4 (8)	0.31 (8)	5.6	5 (-0.6)
Manchester United	0.34 (8)	0.35 (8)	5.5	2 (-3.5)
Wolverhampton Wanderers	0.28 (8)	0.34 (8)	5.0	4 (-1)
Everton	0.32 (8)	0.25 (8)	4.6	3 (-1.6)
Tottenham	0.4 (8)	0.16 (8)	4.5	2 (-2.5)
Watford	0.32 (8)	0.23 (8)	4.3	4 (-0.3)
Sheffield United	0.36 (8)	0.18 (8)	4.3	5 (0.7)
Chelsea	0.42 (8)	0.12 (8)	4.3	3 (-1.3)
Burnley	0.3 (8)	0.22 (8)	4.2	5 (0.8)
Brighton	0.34 (8)	0.18 (8)	4.1	4 (-0.1)
Arsenal	0.16 (8)	0.31 (8)	3.7	2 (-1.7)
Crystal Palace	0.34 (8)	0.12 (8)	3.7	6 (2.3)
Bournemouth	0.24 (8)	0.2 (8)	3.5	0 (-3.5)
Southampton	0.2 (8)	0.2 (8)	3.2	2 (-1.2)
Aston Villa	0.22 (8)	0.09 (8)	2.5	3 (0.5)
Norwich	0.16 (8)	0.15 (8)	2.5	2 (-0.5)
Newcastle United	0.18 (8)	0.11 (8)	2.3	4 (1.7)
West Ham	0.11 (8)	0.07 (8)	1.4	4 (2.6)

optimise team

Optimise your team for a given formation, team value, and assumed bench value. You'll want to have chosen your bench already (there is little point in optimising players on your bench as they will not earn you points a lot of the time).

You can optimise your team based on maximising total points from the historic dataset (full season by default) or points per game ('points' or 'ppg' respectively). If using 'ppg' you will want to set a min_games value so that players with high points per game but with relatively few games played are avoided.

result <- optimise_team(objective = 'ppg', bank = 1000, bench_value = 170, gk = 1, def = 3, mid = 4, fwd = 3, min_games = 3)

Using expected points

If we want to use expected points instead of actual points to optimise the team, we should first fetch the xg data and then we can feed that into the custom_df argument to optimise_team:

df_xp <- fetch_xg_data()

result_xp <- optimise_team(objective = 'ppg', bank = 1000, bench_value = 170, gk = 1, def = 3, mid = 4, fwd = 3, min_games = 3, custom_df = df_xp)

Getting fixtures

We can get and visualise fixtures for the next n games using:

fixtures <- fetch_fixtures(n = 10)

time	gw	team_id	team_name	ha	played	opponent	difficulty	mean_difficulty	median_difficulty
2021-08-13	1	1	Arsenal	a	FALSE	Brentford	2	2.8	2.5
2021-08-14	1	5	Burnley	h	FALSE	Brighton	2	2.9	3.0
2021-08-14	1	4	Brighton	a	FALSE	Burnley	2	2.9	2.5
2021-08-14	1	6	Chelsea	h	FALSE	Crystal Palace	2	2.8	2.0
2021-08-14	1	8	Everton	h	FALSE	Southampton	2	2.7	3.0
2021-08-14	1	9	Leicester	h	FALSE	Wolves	2	2.8	2.5
2021-08-14	1	18	Watford	h	FALSE	Aston Villa	2	2.7	2.5
2021-08-14	1	2	Aston Villa	a	FALSE	Watford	2	3.1	3.5
2021-08-14	1	11	Liverpool	a	FALSE	Norwich	2	2.7	2.0
2021-08-15	1	19	West Ham	a	FALSE	Newcastle	2	2.8	3.0

We can visualise the resulting fixture list:

plot(fixtures)

It may be useful to know teams which have relatively uncorrelated fixtures (good for rotating substitutes):

fixtures <- fixture_rotation(fixtures)

team1	team2	cor
Liverpool	Crystal Palace	0.6755111
Crystal Palace	Leeds	0.6755111
Chelsea	Crystal Palace	0.6805447
Chelsea	Newcastle	0.6998542
Chelsea	Everton	0.7013344
Everton	Liverpool	0.7114582
Everton	Leeds	0.7114582
Burnley	Chelsea	0.7356619
Brighton	Newcastle	0.7419985
Aston Villa	Liverpool	0.7419985