Skip to content
Permalink
master
Switch branches/tags
Go to file
@leesharpe
Latest commit b815d55 May 13, 2020 History
1 contributor

Users who have contributed to this file

Loading

NFL Data Sets

Here is detailed information on each data set.

Draft Picks

To import and join to nflfastR (using the Ben Baldwin name field):

draft_picks <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/draft_picks.csv")
# we do a left join here because the names won't always match but don't want to lose any nflscrapR rows
pbp <- pbp %>%
  left_join(draft_picks,by=c("posteam"="team","name"="name"))

Note this will only match results for the player playing for the team that drafted them. You can remove posteam=team to change this, but the weaker the join, the more likely it results the risk of false positives and duplicate rows for players with names that translate identically once in the NFL play-by-play format, so be careful when you do this! I'd add some additional filtering if you can. For example, if doing a QB analysis, do a filter(position == 'QB') so you're less likely to join to a wrong player.

Data begins with the 2000 season, does not include picks from the supplemental draft, and comes from the excellent Pro Football Reference.

  • season: The season in which the draft occurred. This is the season after the draft, not the one before it, so this would represent the rookie year for drafted players.
  • team: The team that drafted the player. This team may have had this pick originally, or traded for it.
  • round: The round of the draft this pick occurred in.
  • pick: The number of the pick.
  • full_name: The name of the selected player.
  • name: The name of the selected player in the same format as NFL play-by-play data. This will usually match nflscrapR name fields, but not for every player.
  • playerid: The ID of the selected player as used by Pro Football Reference. If NA, the player was not assigned an ID by Pro Football Reference, which normally indicates they never played in an NFL game.
  • side: The side of the ball the player plays on. Can be:
    • For offense, O
    • For defense, D
    • For special teams, S
    • If position is NA, this will be also.
  • category: The category of position the player plays in.
    • For offense, this can be OL, QB, RB, TE, or WR.
    • For defense, this can be DB, DL, ED (edge), or LB. Note that ED is only used from 2019 forward.
    • For special teams, this can be K, KR, LS, P, or ST (generic special teams).
    • If position is NA, this will be also.
  • position: The NFL position the selected player plays as reported by Pro Football Reference. If NA, Pro Football Reference did not record a position.

Draft Values

To import and join to draft pick data from above:

draft_values <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/draft_values.csv")
draft_picks <- draft_picks %>%
  inner_join(draft_values,by=c("pick"="pick"))

Columns:

It's worth noting that the Stuart scale is attempting to measure how teams should value draft picks, while the Johnson and Hill scales are attempting to measure how teams in practice value draft picks, and that these are different questions. The Hill versions is obviously based on more recent data. Also note the systems are using different numerical scales, so you should only compare values within a scale, not compare, say, a Stuart value to a Johnson value (the latter will essentially always be higher).

Games

To import and join to nflfastR data:

games <- read_csv("http://www.habitatring.com/games.csv")
pbp <- pbp %>%
  inner_join(games,by=c("game_id"="game_id","away_team"="away_team","home_team"="home_team"))

Data begins with the 2006 NFL season. Does not include preseason.

Columns:

Colors

To import and join to nflfastR data (for the offense):

Columns:

  • team: The team.
  • team_logo: URL of an image where a transparent team logo is located.
logos <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/teamcolors.csv")
pbp <- pbp %>%
  inner_join(logos,by=c("posteam"="team"))

Columns:

  • team: The team.
  • color: The team's primary color in hexadecimal
  • color2: The team's secondary color in hexadecimal
  • color3: The team's tertiary color in hexadecimal
  • color3: The team's quaternary color in hexadecimal

Special thanks to (@StatsInTheWild)[https://twitter.com/StatsInTheWild] and (@BaumerBen) [https://twitter.com/StatsInTheWild] who did the work of identifying team colors.

Logos

To import and join to nflfastR data (for the offense):

logos <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/logos.csv")
plays <- plays %>%
  inner_join(logos,by=c("posteam"="team"))

This is based off an original version from (Michael Lopez) [https://twitter.com/StatsbyLopez]. Logos mostly come from Wikipedia, except the Tennessee Titans logo image which comes from Hex Sharpe.

Rosters

To import and join to nflfastR (using the Ben Baldwin name field):

rosters <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/rosters.csv")
# we do a left join here because the names won't always match but don't want to lose any nflscrapR rows
plays <- plays %>%
  left_join(rosters,by=c("season"="season,"posteam"="team","name"="name"))

Data begins with the 2006 NFL season and comes from Pro Football Reference.

Columns:

  • season: The season the player was on the roster.
  • team: The team whose roster the player was on.
  • full_name: The name of the selected player.
  • name: The name of the selected player in the same format as NFL play-by-play data. This will usually match nflscrapR name fields, but not for every player.
  • playerid: The ID of the selected player as used by Pro Football Reference. If NA, the player was not assigned an ID by Pro Football Reference, which normally indicates they never played in an NFL game.
  • side: The side of the ball the player plays on. Can be:
    • For offense, O
    • For defense, D
    • For special teams, S
    • If position is NA, this will be also.
  • category: The category of position the player plays in.
    • For offense, this can be OL, QB, RB, TE, or WR.
    • For defense, this can be DB, DL, ED (edge), or LB. Note that ED is only used from 2019 forward.
    • For special teams, this can be K, KR, LS, P, or ST (generic special teams).
    • If position is NA, this will be also.
  • position: The NFL position the selected player plays as reported by Pro Football Reference. If NA, Pro Football Reference did not record a position.
  • games: Number of regular season games the player played in that season.
  • starts: Number of regular season games the player started that season.
  • years: Number of prior seasons of NFL experience the player has. Is 0 for rookies.
  • av: The player's Approximate Value that year, as defined by the Pro Football Reference metric

Standings

To import and join to nflfastR data (for the offense):

standings <- read_csv("http://www.habitatring.com/standings.csv")
plays <- plays %>%
  inner_join(standings,by=c("season"="season","posteam"="team"))

Data begins with the 2002 NFL season.

Columns:

  • season: The year of the NFL season. This reperesents the whole season, so regular season games that happen in January as well as playoff games will occur in the year after this number.
  • conf: The conference the team is in. This will be either AFC or NFC.
  • division: The division the team is in. This will be the value of conf followed by either East, North, South, or West.
  • team: The team.
  • wins: The number of games the team won in the regular season.
  • losses: The number of games the team lost in the regular season.
  • ties: The number of games the team tied in the regular season.
  • pct: The win rate of the team in the regular season. Equals (wins + 0.5 * ties) / (wins + losses + ties).
  • div_rank: This is where this team ranks compared to the other teams in the division based on regular season games only. Will be a number 1-4. If the teams have identical pct values, NFL tiebreakers are applied.
  • scored: The number of points the team has scored in regular season games.
  • allowed: The number of points the team has allowed to be scored on them in regular season games.
  • net: Net points scored in regular season games. Equals scored - allowed.
  • sov: As used in NFL tiebreakers, strength of victory, defined as the combined win rates for teams this team has beaten.
  • sos: As used in NFL tiebreakers, strength of schedule, defined as the combined win rates for teams this team has played.
  • seed: The seed earned by the team in its conference for playoff games. Is NA for teams which do not make the playoffs.
  • playoff: The outcome of the team's playoff run. Is NA for teams which do not make the playoffs, otherwise will be one of LostWC, LostDV, LostCC, LostSB, or WonSB.

Teams

To import:

teams <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/teams.csv")

This data set is designed to help scrape websites for additional NFL information.

Columns:

Trades

To import:

trades <- read_csv("https://raw.githubusercontent.com/leesharpe/nfldata/master/data/trades.csv")

Credit to Pro Football Reference for this information. Data begins with trades occurring 2010-05-01 or later

Columns:

  • trade_id: Arbitrary integer identifying the trade. All rows with the same number are part of the same trade.
  • season: The season in which the trade was executed.
  • trade_date: The date the trade was executed.
  • gave: The team who gave this pick or player up. This is the current abbrevation for this franchise, not the one it had at the time. It consistently favors JAX over JAC, and LAR over LA.
  • received: The team who received this pick or player. This is the current abbrevation for this franchise, not the one it had at the time. It consistently favors JAX over JAC, and LAR over LA.
  • pick_season: The season of the draft in which this traded pick occurs. NA if this row refers to a traded player.
  • pick_round: The round of the draft in which this traded pick occurs. NA if this row referring to a traded player or round number is unknown.
  • pick_number: The pick of the draft in which this traded pick occurs. NA if this row referring to a traded player or pick number is unknown. Note that this may not have been known at the time the trade was executed.
  • conditional: Either 1 if this pick was conditional on some other criteria being met or 0 otherwise. This is a 1 regardless of whether the condition ended up being true.
  • pfr_id: The ID used by Pro Football Reference to refer to the player. If pick_season is NA, this player was part of the trade directly. Otherwise, this player was the player chosen for this pick. Another team may have chosen the player if the pick was subsequently traded.
  • pfr_name: The player's name as shown by Pro Football Reference. If pick_season is NA, this player was part of the trade directly. Otherwise, this player was the player chosen for this pick. Another team may have chosen the player if the pick was subsequently traded.