In this vignette, I’ll walk through how to get started with a basic dynasty value analysis on Sleeper.
We’ll start by loading the packages:
In Sleeper, unlike in other platforms, it’s very unlikely that you’ll remember the league ID - both because most people use the mobile app, and because it happens to be an 18 digit number! It’s a little more natural to start analyses from the username, so let’s start there!
solarpool_leagues <- sleeper_userleagues("solarpool",2020)
head(solarpool_leagues)
#> # A tibble: 3 x 4
#> league_name league_id franchise_name franchise_id
#> <chr> <chr> <chr> <chr>
#> 1 z_dynastyprocess-test 633501761776197~ solarpool 202892038360801~
#> 2 The JanMichaelLarkin Dynasty~ 522458773317046~ solarpool 202892038360801~
#> 3 DLP Dynasty League 521379020332068~ DLP::thoriyan 202892038360801~
Let’s pull the JML league ID from here for analysis, and set up a Sleeper connection object.
jml_id <- solarpool_leagues %>%
filter(league_name == "The JanMichaelLarkin Dynasty League") %>%
pull(league_id)
jml_id # For quick analyses, I'm not above copy-pasting the league ID instead!
#> [1] "522458773317046272"
jml <- sleeper_connect(season = 2020, league_id = jml_id)
jml
#> <Sleeper connection 2020_522458773317046272>
#> List of 5
#> $ platform : chr "Sleeper"
#> $ season : num 2020
#> $ user_name: NULL
#> $ league_id: chr "522458773317046272"
#> $ user_id : NULL
#> - attr(*, "class")= chr "sleeper_conn"
I’ve done this with the sleeper_connect()
function, although you can also do this from the ff_connect()
call - they are equivalent. Most if not all of the remaining functions after this point are prefixed with “ff_”.
Cool! Let’s have a quick look at what this league is like.
jml_summary <- ff_league(jml)
str(jml_summary)
#> tibble [1 x 15] (S3: tbl_df/tbl/data.frame)
#> $ league_id : chr "522458773317046272"
#> $ league_name : chr "The JanMichaelLarkin Dynasty League"
#> $ league_type : chr "dynasty"
#> $ franchise_count: num 12
#> $ qb_type : chr "1QB"
#> $ idp : logi FALSE
#> $ scoring_flags : chr "0.5_ppr"
#> $ best_ball : logi FALSE
#> $ salary_cap : logi FALSE
#> $ player_copies : num 1
#> $ years_active : chr "2019-2020"
#> $ qb_count : chr "1"
#> $ roster_size : int 25
#> $ league_depth : num 300
#> $ prev_league_ids: chr "386236959468675072"
Okay, so it’s the JanMichaelLarkin Dynasty League, it’s a 1QB league with 12 teams, half ppr scoring, and rosters about 300 players.
Let’s grab the rosters now.
jml_rosters <- ff_rosters(jml)
head(jml_rosters)
#> # A tibble: 6 x 7
#> franchise_id franchise_name player_id player_name pos team age
#> <chr> <chr> <chr> <chr> <chr> <chr> <dbl>
#> 1 1 Fake News 1110 T.Y. Hilton WR IND 31.1
#> 2 1 Fake News 1339 Zach Ertz TE PHI 30.1
#> 3 1 Fake News 1426 DeAndre Hopkins WR ARI 28.5
#> 4 1 Fake News 1825 Jarvis Landry WR CLE 28
#> 5 1 Fake News 2025 Albert Wilson WR MIA 28.4
#> 6 1 Fake News 2197 Brandin Cooks WR HOU 27.2
Cool! Let’s pull in some additional context by adding DynastyProcess player values.
player_values <- dp_values("values-players.csv")
# The values are stored by fantasypros ID since that's where the data comes from.
# To join it to our rosters, we'll need playerID mappings.
player_ids <- dp_playerids() %>%
select(sleeper_id,fantasypros_id)
player_values <- player_values %>%
left_join(player_ids, by = c("fp_id" = "fantasypros_id")) %>%
select(sleeper_id,ecr_1qb,ecr_pos,value_1qb)
# Drilling down to just 1QB values and IDs, we'll be joining it onto rosters and don't need the extra stuff
jml_values <- jml_rosters %>%
left_join(player_values, by = c("player_id"="sleeper_id")) %>%
arrange(franchise_id,desc(value_1qb))
head(jml_values)
#> # A tibble: 6 x 10
#> franchise_id franchise_name player_id player_name pos team age ecr_1qb
#> <chr> <chr> <chr> <chr> <chr> <chr> <dbl> <dbl>
#> 1 1 Fake News 1426 DeAndre Ho~ WR ARI 28.5 7
#> 2 1 Fake News 4037 Chris Godw~ WR TB 24.8 16.8
#> 3 1 Fake News 4866 Saquon Bar~ RB NYG 23.8 19.2
#> 4 1 Fake News 4199 Aaron Jones RB GB 26 36.5
#> 5 1 Fake News 4098 Kareem Hunt RB CLE 25.4 56.2
#> 6 1 Fake News 4017 Deshaun Wa~ QB HOU 25.3 59.2
#> # ... with 2 more variables: ecr_pos <dbl>, value_1qb <int>
Let’s do some team summaries now!
value_summary <- jml_values %>%
group_by(franchise_id,franchise_name,pos) %>%
summarise(total_value = sum(value_1qb,na.rm = TRUE)) %>%
ungroup() %>%
group_by(franchise_id,franchise_name) %>%
mutate(team_value = sum(total_value)) %>%
ungroup() %>%
pivot_wider(names_from = pos, values_from = total_value) %>%
arrange(desc(team_value))
value_summary
#> # A tibble: 12 x 8
#> franchise_id franchise_name team_value QB RB TE WR FB
#> <chr> <chr> <int> <int> <int> <int> <int> <int>
#> 1 4 The FANTom Menace 53581 3268 16933 2834 30546 NA
#> 2 3 solarpool 48937 6674 22019 2227 18017 NA
#> 3 1 Fake News 47958 3329 16143 5320 23166 NA
#> 4 11 Permian Panthers 46118 4316 11601 5448 24753 NA
#> 5 12 jaydk 41536 2560 14921 5306 18749 NA
#> 6 8 Hocka Flocka 39549 1940 19032 3202 15375 NA
#> 7 6 sox05syd 36554 3665 3878 7657 21354 NA
#> 8 9 ZPMiller97 32361 3924 10549 3975 13913 NA
#> 9 5 Barbarians 31804 3099 18081 3232 7392 NA
#> 10 7 Flipadelphia05 23730 4192 7452 349 11737 NA
#> 11 2 KingGabe 22640 727 5029 301 16583 NA
#> 12 10 JMLarkin 20887 1113 412 1549 17813 0
So with that, we’ve got a team summary of values! I like applying some context, so let’s turn these into percentages - this helps normalise it to your league environment.
value_summary_pct <- value_summary %>%
mutate_at(c("team_value","QB","RB","WR","TE"),~.x/sum(.x)) %>%
mutate_at(c("team_value","QB","RB","WR","TE"),round, 3)
value_summary_pct
#> # A tibble: 12 x 8
#> franchise_id franchise_name team_value QB RB TE WR FB
#> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <int>
#> 1 4 The FANTom Menace 0.12 0.084 0.116 0.068 0.139 NA
#> 2 3 solarpool 0.11 0.172 0.151 0.054 0.082 NA
#> 3 1 Fake News 0.108 0.086 0.111 0.129 0.106 NA
#> 4 11 Permian Panthers 0.103 0.111 0.079 0.132 0.113 NA
#> 5 12 jaydk 0.093 0.066 0.102 0.128 0.085 NA
#> 6 8 Hocka Flocka 0.089 0.05 0.13 0.077 0.07 NA
#> 7 6 sox05syd 0.082 0.094 0.027 0.185 0.097 NA
#> 8 9 ZPMiller97 0.073 0.101 0.072 0.096 0.063 NA
#> 9 5 Barbarians 0.071 0.08 0.124 0.078 0.034 NA
#> 10 7 Flipadelphia05 0.053 0.108 0.051 0.008 0.053 NA
#> 11 2 KingGabe 0.051 0.019 0.034 0.007 0.076 NA
#> 12 10 JMLarkin 0.047 0.029 0.003 0.037 0.081 0
Armed with a value summary like this, we can see team strengths and weaknesses pretty quickly, and figure out who might be interested in your positional surpluses and who might have a surplus at a position you want to look at.
Another question you might ask: what is the average age of any given team?
I like looking at average age by position, but weighted by dynasty value. This helps give a better idea of age for each team - including who might be looking to offload an older veteran!
age_summary <- jml_values %>%
group_by(franchise_id,pos) %>%
mutate(position_value = sum(value_1qb,na.rm=TRUE)) %>%
ungroup() %>%
mutate(weighted_age = age*value_1qb/position_value,
weighted_age = round(weighted_age, 1)) %>%
group_by(franchise_id,franchise_name,pos) %>%
summarise(count = n(),
age = sum(weighted_age,na.rm = TRUE)) %>%
pivot_wider(names_from = pos,
values_from = c(age,count))
age_summary
#> # A tibble: 12 x 12
#> # Groups: franchise_id, franchise_name [12]
#> franchise_id franchise_name age_QB age_RB age_TE age_WR age_FB count_QB
#> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <int>
#> 1 1 Fake News 27.2 24.9 26.3 27.1 NA 3
#> 2 10 JMLarkin 29.1 25.8 26.6 25.1 0 3
#> 3 11 Permian Panth~ 23.7 23.1 30.9 25.7 NA 3
#> 4 12 jaydk 29.5 25.2 26.1 27.4 NA 4
#> 5 2 KingGabe 25.5 22.1 29.8 21.8 NA 5
#> 6 3 solarpool 25.5 25.2 26.9 27.6 NA 5
#> 7 4 The FANTom Me~ 28.4 23.8 24.3 26.4 NA 4
#> 8 5 Barbarians 25.5 24.1 27.8 26.4 NA 3
#> 9 6 sox05syd 24.1 23.4 26.9 24.3 NA 3
#> 10 7 Flipadelphia05 32.6 25.5 26.6 26 NA 2
#> 11 8 Hocka Flocka 30.3 24.3 24.7 23.5 NA 3
#> 12 9 ZPMiller97 24.4 24.1 26.1 25.1 NA 3
#> # ... with 4 more variables: count_RB <int>, count_TE <int>, count_WR <int>,
#> # count_FB <int>
In this vignette, I’ve used ~three functions: ff_connect, ff_league, and ff_rosters. Now that you’ve gotten this far, why not check out some of the other possibilities?