Sleeper Picks Scraper — Props, Multipliers & Pick Popularity avatar

Sleeper Picks Scraper — Props, Multipliers & Pick Popularity

Pricing

from $0.15 / 1,000 results

Go to Apify Store
Sleeper Picks Scraper — Props, Multipliers & Pick Popularity

Sleeper Picks Scraper — Props, Multipliers & Pick Popularity

Unofficial Sleeper Picks API, no key, no login. Every live player prop line, one row per prop with over and under payout multipliers, odds, pick popularity, recent form, injury status and matchup for NFL, CFB, NBA, NHL, soccer, MMA, tennis, golf, CS2. JSON, CSV, Excel.

Pricing

from $0.15 / 1,000 results

Rating

0.0

(0)

Developer

Logiover

Logiover

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

9 days ago

Last modified

Features

Limited permissions

Share

Not affiliated with Sleeper. This Actor is independent: it is not made, endorsed or sponsored by Sleeper or any league, team or sportsbook. It reads only the board that Sleeper publishes to every visitor and app user — no login, no account, no bets placed. "Sleeper" and every league name belong to their owners and are used here only to say what the data is about.

Sleeper Picks Scraper turns the live Sleeper Picks board into a clean table: one row per prop, with the over and the under side by side — payout multipliers, American odds, implied and no-vig probabilities — plus pick counts and popularity score, the player's last ten games for that stat, injury status, matchup, venue, TV channel and start time. It is a Sleeper Picks API without a key: run it from the Console or call it from code, and get JSON, CSV or Excel for NFL, college football, NBA, WNBA, NHL, soccer (EPL, La Liga, Serie A, Bundesliga, Ligue 1), MMA, tennis, golf and CS2 — whichever leagues Sleeper has lines posted for at that moment. A board of 3,367 lines is 3,367 rows (one per player, stat and game) that you can filter, join and chart without unpacking anything. If you would rather have the over and the under as separate rows with their own ids, flip one switch.

Use it as a player props API, a DFS pick'em scraper, a line shopping feed next to sportsbook odds, a prop model input (implied and no-vig probability, hit rates), an injury monitor, or the start of a line movement and closing line value archive built from scheduled runs. The columns are named like our PrizePicks Scraper (playerName, statType, lineScore, oddsType, startTime, projectionId …), so a Sleeper row and a PrizePicks row can be stacked in one sheet. Export to Google Sheets, Excel, BigQuery, Snowflake, Airtable, n8n, Make or Zapier — and nothing here needs a Sleeper account, a headless browser or a proxy bill.

Table of contents

At a glance

What it readsThe public Sleeper Picks board (all lines, both sides), the promo board, the scoreboard and the player catalogue
Login / API keyNone. No account, no cookies, no browser
LeaguesNFL, CFB, NBA, WNBA, NHL, MLB (when posted), soccer (EPL, La Liga, Serie A, Bundesliga, Ligue 1, MLS), MMA, tennis, golf, CS2 — all that are live
Default outputOne row per line (player × stat × game) with both sides in the row — 102 columns
Other shapesOne row per side (rowShape: side, 115 columns); extra rows for each past game in a line's history (includeRecentGames)
Board size on 2026-10-293,367 lines (= 6,734 sides) in 14 leagues, 1,267 players, 152 games
Biggest proof run34,208 unique rows in 21.2 seconds on 2026-10-29 — rowShape: side plus includeRecentGames (6,734 side rows + 27,474 recent-game rows)
SpeedConsole prefill (100 lines) in 2.7 s, first row after 1.4 s; 1,000 lines in 3 s
FiltersLeague, stat type, player, team, side, multiplier range, minimum picks, promos, start window
PricePay per result: $0.25 per 1,000 rows on the Free plan, $0.15 on Gold and above. Set a maximum cost per run and it stops there
ExportJSON, CSV, Excel, XML, RSS, HTML, or straight into Sheets, Airtable, a webhook or a warehouse
RefreshLines change every few minutes; Sleeper's edge cache keeps a board about 30 seconds

What you get

The board, row by row

A Sleeper "line" is a player, a stat and a game: Aaron Wiggins, Points, 11.5, Cleveland at Atlanta. Sleeper lets you pick the over or the under, and each side carries its own payout multiplier — 1.75 for the over and 1.81 for the under, for example — and its own Sleeper id. The Actor writes the line as one row that holds both sides, then joins everything Sleeper knows about it:

  • The market — league, stat (Points, Pts + Rebs + Asts, Kills (Maps 1+2) …), Sleeper's stat key, the line value, the golf round.
  • The price — over and under multipliers and line ids, each side as American odds, implied and no-vig probability, and the overround.
  • The crowd — picks on the over, the under and in total, each side's share, Sleeper's 0–1 popularity score.
  • The form — the player's last games for that exact stat (up to ten: date, opponent, value), the average, how many cleared the line, fell short or landed on it, and each side's hit rate.
  • The player — name, team, position, headshot, age, college, injury status and note, depth chart slot, ids at Sportradar, ESPN, Yahoo, Rotowire, FantasyData and OddsJam.
  • The game — start time (UTC), opponent, home or away, both teams with records, venue, TV channel, event or tournament, weight class, fixture ids.
  • The promotion — boosted or discounted promo, which kind, whether Sleeper shows it as a free pick.

Everything is a flat column: no nested JSON you have to unpack (the recent-game history is the one array, and it is optional to use because the same facts are also in plain columns).

Three kinds of rows

recordTypeWhenWhat one row is
linedefaultone whole line with both sides in the row (overMultiplier and underMultiplier, overOddsAmerican and underOddsAmerican, the pick counts of both sides …)
siderowShape: sideone side of one line — the unit Sleeper itself prices and ids (lineId) — with the line's columns repeated; twice the rows
recentGameincludeRecentGames: trueone past game from a line's history, with its value and whether it went over, under or pushed against the current line

All three kinds share the unique rowId, so mixing them in one dataset never creates a duplicate, and projectionId / lineKey link a side row or a recent-game row back to its line.

A real row

This is one line row from a platform run on 2026-10-29 (the Console prefill). Empty columns, the external ids and some profile columns are left out here for space; everything shown is as written to the dataset:

{
"rowId": "sleeper:line:1421427865261535255",
"recordType": "line",
"projectionId": "1421427865261535255",
"lineKey": "nba:1393723788809469952:2299:points",
"playerId": "2299",
"playerName": "Aaron Wiggins",
"playerTeam": "ATL",
"playerPosition": "SF",
"jerseyNumber": "2",
"league": "NBA",
"sport": "basketball",
"sleeperSport": "nba",
"statType": "Points",
"statDisplayName": "Points",
"wagerType": "points",
"lineScore": 11.5,
"overMultiplier": 1.75,
"underMultiplier": 1.81,
"overLineId": "1421427865261535255",
"underLineId": "1421427865261535256",
"overOddsAmerican": -133,
"underOddsAmerican": -123,
"overImpliedProbability": 0.5714,
"underImpliedProbability": 0.5525,
"overNoVigProbability": 0.5084,
"underNoVigProbability": 0.4916,
"marketOverround": 0.1239,
"lineType": "normal",
"oddsType": "standard",
"isPromo": false,
"isLive": false,
"gameStatus": "pre_game",
"startTime": "2026-10-30T00:00:00Z",
"updatedAt": "2026-10-29T07:02:03.010Z",
"scrapedAt": "2026-10-29T09:59:39.169Z",
"gameId": "1393723788809469952",
"gameWeek": 2,
"season": "2026",
"seasonType": "regular",
"opponent": "CLE",
"isHome": true,
"homeTeam": "ATL",
"awayTeam": "CLE",
"homeTeamRecord": "2-3",
"awayTeamRecord": "3-1",
"matchup": "CLE @ ATL",
"venueName": "State Farm Arena",
"venueCity": "Atlanta",
"tvChannel": "NBA TV",
"pickCountOver": 214,
"pickCountUnder": 8,
"pickCountTotal": 222,
"pickShareOver": 0.964,
"pickShareUnder": 0.036,
"pickPopularity": 0.9384,
"recentGamesCount": 5,
"recentAverage": 14.6,
"recentOverCount": 4,
"recentUnderCount": 1,
"recentPushCount": 0,
"recentHitRateOver": 0.8,
"recentHitRateUnder": 0.2,
"recentGames": [{"date": "2026-10-21", "opponent": "ORL", "value": 16}, {"date": "2026-10-23", "opponent": "CHA", "value": 19}, {"date": "2026-10-24", "opponent": "HOU", "value": 16}, {"date": "2026-10-26", "opponent": "HOU", "value": 12}, {"date": "2026-10-28", "opponent": "MIA", "value": 10}],
"playerAge": 27,
"depthChartOrder": 2
}

A recent-game row (includeRecentGames) is much shorter — 26 columns: the line's key fields plus pastGameDate, pastOpponent, pastValue, pastResult (over, under or push against the current line) and pastMargin.

Which leagues and stats are on the board

Sleeper decides what is posted, and it changes through the day. On 2026-10-29 the board held these leagues (side rows are two per line; the line count is half of that):

League codeCompetitionSide rowsRecent-game rows
CFBCollege football1,8906,291
NFLNFL1,8528,880
NHLNHL1,3086,391
NBANBA6141,205
CS2Counter-Strike 25522,644
TENNISTennis124620
WNBAWNBA102510
MMAMMA / UFC98107
GOLFGolf78390
LALIGALa Liga (soccer)36161
EPLPremier League (soccer)36128
BUNDESLIGABundesliga (soccer)3094
LIGUE1Ligue 1 (soccer)829
SERIEASerie A (soccer)624

The board had 79 league-and-stat combinations that day. Football lines are dominated by anytime-touchdown and receiving props, basketball by points, rebounds, assists and combinations, hockey by shots, points and assists, esports by kills and headshots over the first two maps, tennis by games, golf by birdies and strokes, MMA by significant strikes and takedowns, and soccer by goalkeeper props (goals against, saves-based fantasy points) with more markets appearing closer to kick-off. Stat keys and labels you have not seen before simply show up in the data: the Actor does not keep a whitelist.

Who uses it and how

1. The DFS pick'em researcher. Sorts the morning board by pickShareOver and recentHitRateOver and shortlists lines where the crowd leans one way and the last ten games the other: {"leagues": ["NBA"], "minPickCount": 100}, then the Pick popularity view.

2. The line shopper. Keeps Sleeper, PrizePicks and sportsbook prices in one workbook. The columns are named alike (playerName, statType, lineScore, startTime), so one lookup finds the same prop everywhere, and oddsAmerican puts Sleeper's multiplier on a sportsbook's scale.

3. The quant building a props model. overImpliedProbability, overNoVigProbability, their under twins and marketOverround give the market's view of both sides; recentAverage, recentOverCount and recentGames give form; injury columns give availability. One run per slate, appended to a warehouse table keyed by projectionId.

4. The newsletter writer. Wants "the five most picked unders tonight": sort the line rows by pickShareUnder (or ask for side: under to get one row per under), join matchup and tvChannel, paste into the template. startTime converts to any time zone.

5. The Discord or Telegram bot developer. Schedules a run every few minutes with startsWithinHours: 3 and onlyPromos: true, diffs projectionId against the last run and posts new boosted or discounted lines. The stable rowId makes de-duplication a set lookup.

6. The fantasy analyst. Cares who is hurt before who is picked: injuryStatus, injuryBodyPart, injuryNotes and depthChartOrder sit next to the line, so a questionable starter with a soft line stands out.

7. The data journalist. Studies crowd behaviour — do users over-pick overs? pickShareOver across thousands of lines, split by league and popularity, answers it in a notebook; daily scheduled runs give a time series without scraping code.

8. The odds-comparison site builder. Needs a feed with ids: projectionId, overLineId, lineKey, gameId, sportradarGameId and oddsjamFixtureId hang Sleeper's pick'em lines next to sportsbook odds for the same fixture.

9. The esports analyst. CS2 lines are kills and headshots over the first two maps; statType, playerTeam, opponent, eventName and up to ten earlier matches in recentGames support map-pool and opponent studies.

10. The tennis, golf and MMA specialist. Tennis rows carry the tournament and opponent for games-played lines, golf rows the tournament and round, MMA rows the event, weightClass and opponent — with the same pick counts as the team sports.

11. The data engineer. Wants a warehouse table: one scheduled run per hour, rows keyed by rowId, a typed all-nullable dataset schema that loads into BigQuery or Snowflake without surprises, and scrapedAt / updatedAt for slowly changing history.

12. The AI-agent builder. Gives an assistant the Actor as a tool (Apify's MCP server or the plain API) so it can answer "what are tonight's most-picked NBA unders and who is injured?" from live data.

Cookbook: ready-made inputs

Paste any of these into the Console's JSON input tab or send them as the run input through the API. Everything not mentioned keeps its default (all leagues, one row per line with both sides, promos included, 20,000 rows maximum).

1. The whole board.

{ "maxItems": 0 }

Every league with lines, one row per line, no row limit. On 2026-10-29 this was 3,367 rows.

2. One league.

{ "leagues": ["NBA"], "maxItems": 0 }

Codes are not case-sensitive: nba, NBA and Nba are the same.

3. Several leagues, one row per side.

{ "leagues": ["NFL", "CFB"], "rowShape": "side", "maxItems": 0 }

The over and the under as separate rows, each with its own lineId, multiplier and pickCount — twice the rows of the default.

4. Find players.

{ "playerNames": ["Jokic", "Doncic", "Lamar"], "maxItems": 0 }

Case and accents are ignored and partial names work. A player with no lines on the board returns nothing.

5. Specific stats.

{ "leagues": ["NBA"], "statTypes": ["Points", "3-Pointers Made", "pts_reb_ast"], "maxItems": 0 }

Use the readable name or Sleeper's key. The run summary lists every stat type on the board if a name matches nothing.

6. Today's games only.

{ "startsWithinHours": 24, "maxItems": 0 }

Or an exact UTC window: { "startDateFrom": "2026-11-01", "startDateTo": "2026-11-01" }.

7. The long shots.

{ "minMultiplier": 2.5, "side": "over", "maxItems": 0 }

Over sides paying at least 2.5x. Setting side switches the output to one row per side; use maxMultiplier for the heavy favourites.

8. What the crowd is picking.

{ "minPickCount": 200, "maxItems": 0 }

Lines with at least 200 picks. Lines Sleeper has no pick statistics for are dropped when this is set.

9. Boosted and discounted lines.

{ "onlyPromos": true }

A handful of rows at a time: discounted lines (oddsType: discount) and boosted multipliers (oddsType: boost).

10. Hit-rate rows.

{ "leagues": ["NBA"], "includeRecentGames": true, "maxItems": 0 }

Adds one recentGame row per past game of every NBA line (up to ten), each marked over, under or push against the current line, with pastMargin. Each of those rows is a billed result.

11. Soccer.

{ "leagues": ["SOCCER"], "maxItems": 0 }

Or one competition: ["EPL"], ["BUNDESLIGA"], ["SERIEA"], ["LALIGA"], ["LIGUE1"], ["MLS"].

12. A small sample of everything.

{ "maxItems": 100 }

When a limit cuts a multi-league board, the rows are a cross-section — every league in turn, soonest games first — rather than only the first league. This is what the Console prefill runs.

Quick start: Console

  1. Open the Actor in the Apify Console and press Try for free (or Start if you already have an account).
  2. Leave the input as it is for a first look: the prefill reads the board and keeps 100 rows, a cross-section of every league that has lines. It finishes in a few seconds.
  3. Open the Dataset tab. The Board view shows league, player, opponent, stat, line, both multipliers and American prices, over pick share, hit rate and start time; other views cover one row per side, probabilities, pick popularity, recent form, injuries, game context and recent-game rows.
  4. Narrow it down: pick leagues in the Leagues box, type players or teams, set Minimum picks on the line, or paste one of the cookbook inputs into the JSON tab.
  5. For the whole board set Max rows to 0 (no limit). Use Maximum cost per run in the run options if you want a hard spending ceiling.
  6. Export the dataset as CSV, Excel, JSON, XML, RSS or an HTML table, or open it in Google Sheets from the same menu.
  7. Use Schedules to run it daily, hourly or every few minutes.

Quick start: API

The Actor id is logiover/sleeper-picks-scraper. Replace <TOKEN> with your Apify API token (Settings, Integrations).

Run and wait for the rows (one call):

curl -s -X POST "https://api.apify.com/v2/acts/logiover~sleeper-picks-scraper/run-sync-get-dataset-items?token=<TOKEN>&format=json" \
-H "Content-Type: application/json" \
-d '{"leagues": ["NBA"], "minPickCount": 50, "maxItems": 0}'

Python (apify-client):

from apify_client import ApifyClient
client = ApifyClient("<TOKEN>")
run = client.actor("logiover/sleeper-picks-scraper").call(run_input={
"leagues": ["NBA", "NHL"],
"minPickCount": 100,
"maxItems": 0,
})
rows = client.dataset(run["defaultDatasetId"]).list_items().items
print(rows[0]["playerName"], rows[0]["statType"], rows[0]["lineScore"], rows[0]["overMultiplier"], rows[0]["underMultiplier"])

Node.js (apify-client):

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: '<TOKEN>' });
const run = await client.actor('logiover/sleeper-picks-scraper').call({ leagues: ['NFL'], statTypes: ['Receiving Yards'], maxItems: 0 });
const { items } = await client.dataset(run.defaultDatasetId).listItems();

The run's key-value store holds a SUMMARY record next to the dataset: the outcome, rows saved by record type and league, lines on the whole board, players and games joined, request routes used, and the stat types on the board (GET /v2/key-value-stores/<STORE_ID>/records/SUMMARY).

Input reference

InputTypeDefaultWhat it does
leaguesarray of textempty = every league with linesLeague codes: NFL, CFB, NBA, WNBA, NHL, MLB, MMA, TENNIS, GOLF, CS2, EPL, LALIGA, SERIEA, BUNDESLIGA, LIGUE1, MLS, or a family (SOCCER, FOOTBALL, BASKETBALL, HOCKEY, ESPORTS). Aliases UFC, PGA, CS, NCAAF work; so does Sleeper's raw code (clubsoccer:epl). Case does not matter.
statTypesarray of textempty = allReadable stat names (Points, Passing Yards) or Sleeper's keys (pts_reb_ast, anytime_touchdowns). Case, spaces and punctuation are ignored.
playerNamesarray of textempty = allPartial names; accents and case ignored.
teamsarray of textempty = allTeam codes (KC, LAL) or, where available, names. Matches the player's team and either side of the matchup.
sideboth / over / underbothKeep one side only. Choosing one switches the output to one row per side.
minMultiplier, maxMultipliernumbernoneKeep sides paying at least / at most this multiplier (line rows: either side).
minPickCountintegernoneKeep lines with at least this many picks in total. Lines without pick statistics are dropped when set.
includePromosyes / noyesAdd Sleeper's boosted and discounted promo lines (a handful) and flag them.
onlyPromosyes / nonoOnly the promo lines.
startsWithinHoursinteger 1–9,600noneGames starting within this many hours from now.
startDateFrom, startDateToYYYY-MM-DDnoneGames starting on or after / on or before this UTC day.
rowShapeline / sidelineOne row per line with both sides in it (default), or one row per side (twice the rows).
includeRecentGamesyes / nonoAdd one row per past game in each line's history. Each is a billed result.
maxItemsinteger 0–1,000,00020,000 (prefill 100)Rows saved at most; 0 = no limit.
proxyConfigurationproxyno proxyForce a proxy for every request. Normally not needed.

All filters combine with AND. Invalid values stop the run with INVALID_INPUT and a message that names the field.

Data dictionary

Every column, with its type, meaning, the measured fill rate and an example from a real row. Rows: L = line rows (the default), S = side rows (rowShape: side), R = recent-game rows. Fill rates were measured over the 2026-10-29 proof run (6,734 side rows, 27,474 recent-game rows): both sides of a line carry the same line data, so every line-level column fills at the same rate on a line row; the over/under odds and probability columns are derived from the multipliers and fill whenever both multipliers do. Where a column is in both S and R the two rates are shown as side / recent-game. A fill rate below 100% is not a defect: the source does not carry the value for those rows (a golf line has no opponent, a tennis player no team, a line nobody picked no pick statistics).

Identity

FieldType · rowsMeaningFilledExample
rowIdtext · SLRUnique key of the row100% / 100%sleeper:line:1421427865…
recordTypetext · SLRside, line or recentGame (see Three kinds of rows)100% / 100%line
platformtext · SLRAlways sleeper100% / 100%sleeper
projectionIdtext · SLRThe over side's line id, shared by both sides of a line100% / 100%1421427865261535255
lineKeytext · SLR<sleeperSport>:<gameId>:<playerId>:<wagerType> (+ round / promo suffix); pairs the sides100% / 100%nba:1393723788809469952…
marketTypetext · SLSleeper's own market string, for example points_over_under_player_1240100%points_over_under_playe…

Player

FieldType · rowsMeaningFilledExample
playerIdtext · SLRSleeper player id100% / 100%2299
playerNametext · SLRFull name from Sleeper's player catalogue100% / 100%Aaron Wiggins
playerTeamtext · SLRTeam code (GB, CLE) or club name in soccer; empty in tennis, MMA, golf95.5% / 95.9%ATL
playerTeamNametext · SLTeam nickname when the scoreboard carries one49.2%
playerPositiontext · SLRPosition on the line (QB, WR, PG, C, D)87.3% / 86.3%SF
positionRanktext · SLRank within the team's position group, football only (RB2, WR1)27.5%
jerseyNumbertext · SLJersey number87.3%2
playerImageUrltext · SLSleeper headshot where one exists60.9%https://sleepercdn.com/…

Market

FieldType · rowsMeaningFilledExample
leaguetext · SLRNFL, CFB, NBA, WNBA, NHL, MMA, TENNIS, GOLF, CS2, EPL, LALIGA …100% / 100%NBA
sporttext · SLRFamily: football, basketball, hockey, soccer, mma, tennis, golf, esports100% / 100%basketball
sleeperSporttext · SLRSleeper's sport code (nfl, nba, cs, clubsoccer:epl …)100% / 100%nba
statType, statDisplayNametext · SLRReadable stat name (Points, Pts + Rebs + Asts, Passing Yards); the second column repeats it under the PrizePicks Scraper's name100% / 100%Points / Points
wagerTypetext · SLRSleeper's stat key (points, pts_reb_ast, kills_maps_1_2 …)100% / 100%points
lineScorenumber · SLRThe over/under threshold (the line)100% / 100%11.5
roundnumber · SLGolf only: the tournament round the line is for1.2%

Side & payout

FieldType · rowsMeaningFilledExample
sidetext · Sover or under100%over
lineIdtext · SSleeper's id of this side (globally unique)100%1421442837966237716
oppositeLineIdtext · SSleeper's id of the other side of the same line100%1421442837966237717
multiplier, oppositeMultipliernumber · SPayout multiplier of this side and of the other side100%1.95 / 1.64
oddsDecimal, oddsAmericannumber · SThe multiplier read as decimal odds, and as American odds (1.78 is -128)100%1.95 / -105
impliedProbability, noVigProbabilitynumber · S1 / multiplier (break-even win rate), and the same with the margin removed so the pair sums to 1100%0.5128 / 0.4568
overMultiplier, underMultipliernumber · SLPayout multiplier of each side (both row shapes)100%1.75 / 1.81
overLineId, underLineIdtext · SLSleeper line id of each side100%1421427865261535255 / 1…
overOddsAmerican, underOddsAmericannumber · SLEach side's multiplier as American odds (1.78 is -128); the decimal price is the multiplier100%-133 / -123
overImpliedProbability, underImpliedProbabilitynumber · SL1 / multiplier for each side: the break-even win rate100%0.5714 / 0.5525
overNoVigProbability, underNoVigProbabilitynumber · SLEach side's implied probability with the margin removed (the pair sums to 1)100%0.5084 / 0.4916
marketOverroundnumber · SLImplied over + implied under - 1: the margin in the pair (about 0.12)100%0.1239

Promotion

FieldType · rowsMeaningFilledExample
lineType, oddsTypetext · SLSleeper's line type (normal, line_promotion) and the tier name (standard, discount, boost)100%normal / standard
isPromo, promotionType, displayAsFreePickyes/no · SLPromotional-line flag, Sleeper's promotion label, and whether it is shown as a free pick100% / 0.1% / 0.1%False

Status & time

FieldType · rowsMeaningFilledExample
statustext · SLLine status at Sleeper (active)100%active
isLiveyes/no · SLTrue when the game has started100%False
gameStatustext · SLGame state on the line: pre_game100%pre_game
startTimetext · SLRGame start, ISO-8601 UTC100% / 100%2026-10-30T00:00:00Z
updatedAttext · SLWhen Sleeper last changed the line, ISO-8601 UTC100%2026-10-29T07:02:03.010Z
scrapedAttext · SLRWhen the run built the row, ISO-8601 UTC100% / 100%2026-10-29T09:59:39.169Z

Game

FieldType · rowsMeaningFilledExample
gameIdtext · SLRSleeper game id (a number, 20261029_DAL_MTL style, or match-<n>)100% / 100%1393723788809469952
gameWeeknumber · SLSleeper's week of the season (week of the year in individual sports)100%2
seasontext · SLSeason year100%2026
seasonTypetext · SLregular or post100%regular
opponenttext · SLOpponent team code, or the other competitor in tennis and MMA98.8%CLE
isHomeyes/no · SLPlayer's side is the listed home slot (order is arbitrary in individual sports)98.8%True
homeTeam, awayTeamtext · SLHome and away team codes (club names in soccer without a code); the two competitors in tennis, MMA, CS298.8%ATL / CLE
homeTeamName, awayTeamNametext · SLTeam nicknames where the scoreboard carries them60.7%
homeTeamRecord, awayTeamRecordtext · SLTeam records going into the game (4-3-0, 2-3)87.3%2-3 / 3-1
matchuptext · SLCAR @ GB in team sports, A vs B in tennis, MMA, esports98.8%CLE @ ATL
venueName, venueCity, venueState, venueCountrytext · SLVenue name, city, state or province, and country87.3% / 87.3% / 85.8% / 87.3%State Farm Arena / Atla…
tvChanneltext · SLBroadcaster Sleeper lists for the game (ESPN, AMZN, Peacock)79.9%NBA TV
eventNametext · SLEvent title (Thursday Night Football, UFC Fight Night: …)17.5%
tournamenttext · SLTournament name (tennis and golf)3%
matchTypetext · SLTennis match type (women_singles)1.8%
weightClasstext · SLMMA weight class1.5%
sportradarGameId, oddsjamFixtureIdtext · SLGame / fixture ids at Sportradar and OddsJam, for joining other data88.2% / 36.6%8202c938-cdf3-4e4e-8f88…

Pick popularity

FieldType · rowsMeaningFilledExample
pickCount, pickSharenumber · SPicks on this side, and its share of all picks on the line (0-1)58.8%6 / 0.6667
pickCountOver, pickCountUnder, pickCountTotalnumber · SLPicks on the over, on the under, and in total58.8%214 / 8 / 222
pickShareOver, pickShareUndernumber · SLOver and under share of the picks (0-1)58.8%0.964 / 0.036
pickPopularitynumber · SLSleeper's own popularity score for the line (0-1, higher is more picked)58.8%0.9384

Recent form

FieldType · rowsMeaningFilledExample
recentGamesCountnumber · SLRNumber of past games in the line's history (up to 10)99.8% / 100%5
recentAveragenumber · SLAverage of the stat over those games99.8%14.6
recentOverCount, recentUnderCount, recentPushCountnumber · SLPast games above, below and exactly on the current line99.8%4 / 1 / 0
recentHitRateOver, recentHitRateUndernumber · SLThose counts divided by the number of past games (4 decimals)99.8%0.8 / 0.2
recentHits, recentHitRatenumber · SPast games this side would have won, and that divided by the games99.8%2 / 0.5
recentGameslist · SLThe history itself, oldest first: [{date, opponent, value}]99.8%

Player profile

FieldType · rowsMeaningFilledExample
playerAgenumber · SLAge42.8%27
playerHeight, playerWeighttext · SLHeight in inches and weight in pounds, as Sleeper stores them89.3% / 88.9%77 / 190
playerCollege, playerBirthDate, playerBirthCountry, playerYearsExptext · SLCollege, birth date, birth country, years of experience37.4% / 62.3% / 4.6% / 85%Maryland / 1999-01-02 / 5
playerStatustext · SLRoster status in Sleeper's catalogue (Active, ACT, active)90.3%ACT
injuryStatus, injuryBodyPart, injuryNotestext · SLInjury designation (Questionable, Out, IR), body part and note; empty when healthy6.5% / 6.4% / 0.6%
depthChartPosition, depthChartOrdertext · SLDepth chart slot and order (1 = starter)36.6%PF / 2
fantasyPositionslist · SLPositions the player is eligible for in fantasy formats39.9%

External ids

FieldType · rowsMeaningFilledExample
sportradarId, espnId, yahooId, rotowireId, fantasyDataId, oddsjamIdtext · SLPlayer ids at Sportradar, ESPN, Yahoo, Rotowire, FantasyData and OddsJam90.1% / 9.5% / 9.5% / 85.3% / 36.6% / 33.4%d3b28775-473d-471a-8452…

Recent-game rows

FieldType · rowsMeaningFilledExample
recentGameNumbernumber · RPosition of the game in the history, 1 = oldest100%4
pastGameDate, pastOpponent, pastValuetext · RDate, opponent and stat value of the past game100% / 96.3% / 100%2026-01-04 / DAL / 3
pastResult, pastMargintext · Rover, under or push against the current line, and the value minus the line100%over / 0.5

Which leagues fill which columns

playerAge is empty for hockey, college football, MMA and CS2 because Sleeper's catalogue carries no age for most of those players (hockey rows usually still have playerBirthDate); oddsjamId and espnId exist mainly for NFL and NBA players; tvChannel and venueName are published for most team-sport games but not for the individual sports; opponent is empty only in golf. Nothing is invented to fill a gap.

Conditional columns

These columns are empty on most rows by design; each one fills completely in the leagues where Sleeper publishes the value (measured in the proof run, share of rows within those leagues):

ColumnWhat it holdsFills in (share within the league)All rows
homeTeamName, awayTeamNameteam nicknames and club names (hockey, college football, soccer), the competitors' names in tennis, MMA and CS2BUNDESLIGA, CFB, CS2, EPL, LALIGA, LIGUE1, MMA, NHL, SERIEA, TENNIS (100%)60.7%
playerTeamNamethe player's team nickname or club name where the scoreboard has oneBUNDESLIGA, CFB, EPL, LALIGA, LIGUE1, NHL, SERIEA (100%)49.2%
matchTypetennis match type (women_singles)TENNIS (100%)1.8%
playerBirthCountryplayers whose catalogue entry has a birth country (soccer, tennis, golf)BUNDESLIGA 100%, EPL 94.4%, GOLF 100%, LALIGA 100%, LIGUE1 100%, SERIEA 100%, TENNIS 93.5%4.6%
roundgolf: the tournament round of the lineGOLF (100%)1.2%
tournamenttennis and golf: the tournament nameGOLF, TENNIS (100%)3%
weightClassMMA: the fight's weight classMMA (100%)1.5%

CSV columns

Line rows (the default) have 102 columns — every field marked L above. Side rows have 115: the same plus the per-side columns (side, lineId, oppositeLineId, multiplier, oppositeMultiplier, oddsDecimal, oddsAmerican, impliedProbability, noVigProbability, pickCount, pickShare, recentHits, recentHitRate). Recent-game rows have 26. A line row never carries empty per-side columns: the keys are simply absent. In a CSV export recentGames and fantasyPositions are JSON text inside the cell; in JSON they are real arrays.

How the numbers are computed

  • overImpliedProbability = 1 ÷ overMultiplier to four decimals (the under twin likewise): 1.78 implies 0.5618, the win rate a single pick needs to break even. overNoVigProbability divides it by the sum for both sides, so the pair adds to 1; marketOverround is that sum minus 1 (median 0.1237 in the proof run, range 0.0653 to 0.1594). On side rows the same numbers for the row's own side are impliedProbability and noVigProbability.
  • overOddsAmerican converts the multiplier as a decimal price: 1.78 is −128, 2.36 is +136, 3.5 is +250. The decimal price is the multiplier itself.
  • pickShareOver = picks on the over ÷ all picks on the line. recentOverCount counts past games strictly above lineScore, recentUnderCount strictly below, recentPushCount exactly on it; hit rates divide by the past games, so ties lower both. recentAverage is the mean of the listed games.

These convert numbers Sleeper publishes. A multiplier is a pick'em payout factor; reading it as a decimal price is a convenience for comparison with sportsbook odds, not a statement about how Sleeper settles an entry — how picks combine into an entry payout is Sleeper's own rule, which this Actor neither reads nor reproduces.

Scale, speed and measurements

All numbers below were measured on the Apify platform on 2026-10-29 with the default 1,024 MB memory setting.

RunInputResult
Console prefill{"maxItems": 100} (one row per line, the default)100 rows, finished in 2.7 s, first row after 1.4 s
1,000 lines{"maxItems": 1000}1,000 rows in 3 s
The whole board, one row per side{"rowShape": "side", "maxItems": 0}6,734 rows in 9.2 s (the default shape writes half as many rows)
Proof run{"rowShape": "side", "maxItems": 0, "includeRecentGames": true}34,208 rows, 34,208 unique rowIds, 0 duplicates, 21.2 s (6,734 side rows + 27,474 recent-game rows)

Which shape the proof used. The ≥10,000-row proof ran with rowShape: side and includeRecentGames: true on build 1.0.4, the final extraction code. Later builds changed only the default row shape (now one row per line), added the over/under American-odds and probability columns, the budget-stop message and this README; the fetching and joining are unchanged.

Why the proof run adds recent-game rows. The whole public board is smaller than 10,000 rows: 3,367 lines are 6,734 sides on 2026-10-29, and the Actor returns every line and side on it. Each line also carries up to ten past games, and includeRecentGames writes each as its own row — real data, which is how one run reaches 34,208 rows. Without it, expect the board's own size: 3,367 rows by default, 6,734 with rowShape: side.

A run makes about 40–60 small requests (the board and the promo board, then per sport the player catalogue, the current week and its scoreboard), almost all straight from the platform. The board is one cached JSON of about 5.7 MB (390 KB compressed). Most of the time goes into writing rows; the Actor writes in batches as soon as the data is joined and stops cleanly at a time or cost limit.

Connection, measured on 2026-10-29: Sleeper's public endpoints answered a direct connection from the platform, the Apify datacenter proxy, a US residential exit and a laptop in Turkey; no geographic block was seen. The Actor goes direct first, retries once, then uses the datacenter proxy, and at most eight residential requests per run as a last resort. A run that cannot read the board fails with BLOCKED or FETCH_FAILED; it never reports an empty result for an access problem.

What you pay

Pay per result. Every row written to the dataset is one result — by default one line, one prop. There is no start fee and no charge for the summary record.

PlanPrice per 1,000 rows
Free$0.25
Bronze$0.217
Silver$0.183
Gold, Platinum, Diamond$0.15
RowsFree planGold plan and above
1,000$0.25$0.15
10,000$2.50$1.50
100,000$25.00$15.00

Today's whole board in the default shape (3,367 lines, 3,367 rows) costs about $0.84 on the Free plan and about $0.51 on Gold and above. rowShape: side doubles the rows (6,734: about $1.68 and $1.01), and leagues, startsWithinHours or minPickCount shrink it. Optional recent-game rows are billed like any other row — leave them off unless you want the flat game-log table.

Maximum cost per run. Apify lets you set a spending ceiling on every run of a pay-per-result Actor (run options, "Maximum cost per run"). The Actor reads it before it writes anything and never writes past it: the dataset ends exactly at what the ceiling pays for, the run finishes as succeeded, and the status message says so — for example "Saved 230 rows — this run's maximum cost ($0.05) covers 230 results at the current price. Raise "Maximum cost per run" to get more." (the number depends on your plan's price per result: $0.05 buys 200 rows on the Free plan). Rows already written are kept and billed; nothing is written and then cut. If the ceiling covers no result at all the run says that too.

Integrations and scheduling

  • Schedules. In the Console, Schedules runs the Actor on a cron: daily for a morning board, hourly for movement, every few minutes on game day. Sleeper's edge cache holds the board for about 30 seconds (the response header says s-maxage=30), so polling faster than every minute mostly re-reads the same snapshot.
  • Webhooks. A webhook on ACTOR.RUN.SUCCEEDED can call your endpoint with {"runId": "{{resource.id}}", "datasetId": "{{resource.defaultDatasetId}}"}; fetch datasets/<id>/items from there.
  • Google Sheets. The Apify integration appends a finished run's dataset to a sheet; point it at the Board columns or use fields= in the dataset URL to keep it narrow.
  • Airtable, Make, Zapier, n8n. Use the Run Actor module, then Get dataset items. Filter on isPromo, pickShareOver or overOddsAmerican inside the scenario, or push the filters into the Actor's input so you pay for fewer rows.
  • BigQuery, Snowflake, Postgres. Export JSON or CSV from the dataset API on a schedule and load it; the dataset schema is typed, every column is nullable, and rowId is the primary key. For history, append rows and keep scrapedAt.
  • MCP and AI agents. Apify's MCP server exposes any Actor as a tool; give an assistant logiover/sleeper-picks-scraper with a narrow input (leagues, startsWithinHours, maxItems) so answers come from live data.
  • Line-movement archive. Schedule a run, append to a table, and compare consecutive snapshots on lineKey: a change in lineScore is a moved line, a change in a multiplier is a repriced side, a new lineKey is a new line, a missing one is a pulled line. The Actor does not keep history for you — your schedule does, and the dated rows (scrapedAt, updatedAt) make the table self-explanatory.

Performance tips

  1. Ask only for what you need: leagues, startsWithinHours and minPickCount cut rows before they are billed, and the default one-row-per-line shape is already half of rowShape: side.
  2. Leave includeRecentGames off unless you want the flat game-log table; the same history is in the recentGames column of every normal row.
  3. Do not poll faster than once a minute, and leave the proxy empty: the direct route was the fastest in every measurement.
  4. Key on projectionId (lines) or lineId (sides) when you de-duplicate or diff runs, and set Maximum cost per run while you experiment.

Troubleshooting

The Actor ends with one of these outcomes in the status line and in the SUMMARY record.

Message (starts with)MeaningWhat to do
COMPLETED: saved N rows …Normal finish. May continue with "the maxItems limit of N was reached" or "this run's maximum cost of $X covers N results".Raise maxItems / the cost ceiling if you wanted more.
PARTIAL: saved N rows …Rows were saved, but the time limit was reached, or a player catalogue / scoreboard request failed ("no names for those rows", "no start times for those rows").Run again; the board is public and the failure is usually momentary.
EMPTY_RESULT: Sleeper has no lines posted for … right nowThe league you asked for has no lines (off-season, or no games soon).Leave leagues empty to see every league, or try later.
EMPTY_RESULT: Sleeper Picks has no lines on the board right nowThe whole board is empty.Try later.
EMPTY_RESULT: the board has N lines in … but none match these filtersYour filters exclude everything. The message lists leagues and stat types that exist.Loosen a filter; check spelling of a player or stat.
INVALID_INPUT: …A value is out of range or malformed; the message names the field (rowShape, maxItems, minMultiplier, maxMultiplier, minPickCount, startsWithinHours, startDateFrom, startDateTo).Fix the field.
BLOCKED: no board collected … / FETCH_FAILED: no board collected …The board could not be read after the retries (an access problem or a network failure — never an empty result).Run again; set a proxy in the input if the failure repeats.
"does not cover a single result"The maximum cost per run is lower than the price of one row.Raise it.

SUMMARY.notes records anything partial: players missing from Sleeper's catalogue (no name or profile), games missing from the scoreboard (no start time or venue), an unreadable promo board, a restart that resumed an earlier run, a side filter that switched the shape, and cross-section sampling when a limit cut the board.

Honest limits

  • A snapshot, not a tape. Each run returns the board as it is now: no tick history, no settled results. Schedule runs to build your own history.
  • Pre-game only. Every line on the board was pre-game on 2026-10-29; Sleeper removes lines when games start, and there is no in-game feed.
  • One line per player, stat and game. Sleeper's public board carried no alternate-line ladders on 2026-10-29: apart from a handful of promos there is one line per player, stat and game. If Sleeper adds alternates, the Actor passes them through as extra lines, because it keeps every line the board lists.
  • Pick statistics are partial. 58.8% of side rows had pick counts; Sleeper publishes none for a line nobody has picked. The 0–1 popularity score is Sleeper's own number and its formula is not published.
  • The board follows the calendar. MLB is dark in the off-season and soccer markets arrive near kick-off; a league with nothing posted returns nothing, and that is reported, not an error.
  • Individual sports have no team. Tennis, MMA and golf rows have no team, venue or TV channel; homeTeam / awayTeam hold the two competitors in Sleeper's order. Names are Sleeper's spelling; use the external ids to join other sources.
  • Multipliers are not sportsbook odds. oddsDecimal, oddsAmerican and impliedProbability are arithmetic on Sleeper's multiplier — useful for comparison and modelling, not a statement of how any entry is paid.
  • Availability differs by place. The public board is the same for everyone who can read it; whether you may play Sleeper Picks or use betting data in your country is a legal question this Actor does not decide.

Comparison with the other ways to get this data

WayGood forFalls short
Looking in the Sleeper appOne player, right nowNo export, no history, no cross-league filters
Sleeper's documented APILeagues, rosters, drafts, playersBuilt for fantasy leagues; as far as we can tell it does not document the Picks board
Your own script on the public JSONFull controlYou must join four sources, handle sport-code quirks (soccer competitions, split-year seasons, leagues whose "current week" lags), keep it running and fix it when fields change
This ActorThe board as clean, joined, documented rows on demand or on a scheduleSnapshot only, pre-game only; what Sleeper does not publish is not there

Frequently asked questions

Is there a Sleeper Picks API?

Sleeper documents an API for fantasy leagues, rosters and drafts, but as far as we can tell not the Picks board. The board is public JSON that Sleeper's own apps read without a login; this Actor returns it as clean rows — a Sleeper Picks API without a key or scraping code.

How do I scrape Sleeper Picks player props?

Run the Actor with {"maxItems": 0} for every league on the board, or add leagues, playerNames or statTypes. Each row is one line (one prop) with both payout multipliers, odds, pick counts, recent form and injury status. Export JSON, CSV or Excel.

Do I need a Sleeper account, token or login?

No. The Actor reads only what any visitor or app user can see; it never logs in and never places a pick.

Which sports and leagues does it cover?

Whatever Sleeper has lines for: NFL, college football, NBA, WNBA, NHL, MMA, tennis, golf, CS2 and, when posted, soccer (EPL, La Liga, Serie A, Bundesliga, Ligue 1, MLS) and MLB — 14 leagues on 2026-10-29. A league with no lines returns nothing and the status message says so.

What is a payout multiplier on Sleeper Picks?

The factor Sleeper posts for a side of a line; a 1.78 side pays 1.78 times the stake when played on its own and wins. How several picks combine into an entry payout is defined by Sleeper's rules, not by this Actor.

How do I turn a Sleeper multiplier into implied probability?

Divide 1 by the multiplier — overImpliedProbability and underImpliedProbability do it to four decimals, and the NoVig twins remove the margin so the two sides sum to 1. A 1.78 / 1.78 line implies 0.5618 each, a 12% overround.

What is the difference between line rows and side rows?

A line row (the default) holds the over and the under of one prop side by side: overMultiplier / underMultiplier, the American odds, probabilities and pick counts of both. A side row (rowShape: side) is the over or the under on its own with its own Sleeper id (lineId) and costs twice as much, because there are two per line.

What does pickPopularity mean, and why is it sometimes empty?

It is Sleeper's own 0–1 popularity score; how it is computed is not published. Sleeper attaches pick statistics only to lines that have been picked — 58.8% of side rows on 2026-10-29 — so empty means "none published", not zero. For exact crowd behaviour use the counts and shares.

How often do Sleeper lines change?

Every few minutes on a busy day (updatedAt shows the last change). Sleeper's edge cache serves a board for about 30 seconds, so scheduling faster than once a minute re-reads the same snapshot.

Does it include alternate lines?

Not on 2026-10-29: Sleeper's public board carried one line per player, stat and game plus a handful of promos. The Actor keeps every line the board lists, so alternates would appear as extra lines with their own lineId and lineScore if Sleeper published them.

What are promo, boosted and discounted lines?

A few promotional lines Sleeper posts: a discounted threshold (oddsType: discount, often a free pick) or a boosted multiplier (oddsType: boost). They are flagged with isPromo, promotionType and displayAsFreePick; onlyPromos returns just those.

Does it include live, in-game lines?

No. Every line on the board was pre-game on 2026-10-29; Sleeper removes lines when a game starts. isLive and gameStatus are on every row in case that changes.

Can I get historical Sleeper lines, line movement or closing line value?

Sleeper keeps no history on the public board. Schedule the Actor, append each run to a table keyed by lineKey and scrapedAt, and compare snapshots: a changed lineScore is a moved line, a changed multiplier a repriced side, and the last snapshot before startTime is your closing line.

Can I compare Sleeper with PrizePicks, Underdog or sportsbook odds?

Yes — the column names follow the PrizePicks Scraper, and sportradarId, oddsjamId, sportradarGameId and oddsjamFixtureId identify the same player and game in other sources. A pick'em multiplier is not a sportsbook price, so convert with care.

Is there arbitrage or +EV between Sleeper and sportsbooks?

Pick'em lines and sportsbook props have different payout rules, so a gap between a Sleeper implied probability and a book's no-vig price is a research signal, not a guaranteed profit. The Actor supplies clean inputs (overNoVigProbability, recentHitRateOver); estimating the true probability is your model's job.

How do I find the most picked players tonight?

{"startsWithinHours": 24, "minPickCount": 100, "maxItems": 0}, then sort the dataset by pickCountTotal or pickShareOver.

How do I filter by player, team or stat?

playerNames, teams and statTypes are lists. Names ignore case and accents and match partial text; teams match by code (and by name where Sleeper lists one) on either side of the matchup; stats match the readable name or Sleeper's key.

Why does my player filter return nothing?

The player probably has no line on the board right now (injured, resting, not in a covered game), or the name is spelled differently — try the surname. The status message lists the leagues and stat types the board has.

What do wagerType and statType mean?

wagerType is Sleeper's key (pts_reb_ast, kills_maps_1_2); statType is the readable label (Pts + Rebs + Asts, Kills (Maps 1+2)). A new Sleeper stat appears under a readable version of its key until a label is added.

What do gameWeek and isHome mean?

gameWeek is Sleeper's week number for the game (week of the season in football, the week of the year in tennis, golf, MMA and CS2). isHome says which slot the player occupies; in individual sports and esports the home/away order carries no meaning.

Why do tennis, MMA and golf rows have no team or venue?

They are individual sports: Sleeper has no team and, for most events, no venue or TV channel. The columns stay empty rather than guessed; tournament, event name and weight class are filled where Sleeper gives them.

How do I get the player's recent games?

Every normal row has recentGames (date, opponent, value; up to ten) for that exact stat, plus the average, over/under/push counts and hit rates. includeRecentGames writes each past game as its own row instead.

Does it give injury information and headshots?

Injury status, body part, note and depth chart come from Sleeper's player catalogue; most players have none (436 of 6,734 side rows in the proof run carried a status). playerImageUrl is Sleeper's headshot for NFL, NBA, WNBA, NHL, tennis and MMA players; other sports have no image.

How much does a full run cost, and how do I cap it?

The whole board in the default shape is 3,367 rows: about $0.84 on the Free plan and $0.51 on Gold and above ($0.25 / $0.15 per 1,000 rows); rowShape: side doubles it. Set Maximum cost per run in the run options: the Actor writes only what the ceiling pays for, finishes successfully and says so.

Can I run it from outside the US?

Yes. It runs on the Apify platform, and the endpoints also answered from Turkey without a geographic block. Whether you may play Sleeper Picks or use betting data where you live is a separate legal question.

The Actor reads public data served to every visitor, without logging in or bypassing any access control. Whether a particular use is permitted depends on Sleeper's terms, your jurisdiction and your purpose — see the legal chapter; you are responsible for your own compliance.

Can I use the data to place picks automatically?

No. The Actor never logs in to Sleeper and cannot place, change or cancel a pick; it supplies data for research.

Which export formats and integrations work?

JSON, CSV, Excel, XML, RSS and HTML from the dataset's Export menu or the API; Google Sheets, Airtable, n8n, Make, Zapier, webhooks and Apify's MCP server for AI assistants.

How do I avoid duplicates between runs?

Use rowId (or projectionId for lines, lineId for sides). A line keeps its Sleeper id while it is on the board; if Sleeper reprices a side the id stays and the multiplier changes — that is how you detect movement.

What happens if Sleeper changes something?

Unknown stats appear under a readable version of their key, unknown leagues under their uppercase code, and a missing player or game leaves dependent columns empty and is reported in the summary. If Sleeper restructures an endpoint, runs fail with a clear message rather than return wrong data, and we update the Actor.

Who do I contact about a bug or a missing column?

Open an issue on the Actor's page or message us — we reply within 24 hours.

Glossary

  • Pick'em — picking the over or under on player stat lines instead of betting at fixed odds.
  • Line / side — a player, a stat and a game with a threshold (Points 11.5); each line has an over side and an under side, with its own Sleeper id and multiplier. Landing exactly on the threshold is a push.
  • Payout multiplier — the factor Sleeper posts for a side. Implied probability — 1 ÷ multiplier, the win rate that breaks even. No-vig probability — that, with the margin removed. Overround (vig, hold) — the margin: the implied probabilities' sum minus 1.
  • Hit rate — the share of recent games that cleared a threshold. Pick count / share — how many users picked a side and its share of the line's picks. Popularity score — Sleeper's own 0–1 ranking number.
  • Promo, boost, discount, free pick — Sleeper's promotional lines: a raised multiplier, or a lowered threshold shown at no cost.
  • Wager type — Sleeper's key for the stat (passing_yards). Board — all lines open at a moment.
  • Line shopping, line movement, closing line value, +EV, arbitrage — comparing a prop across platforms; how a line changes after you take it; how your number compares with the last one before the game; a pick whose true win probability beats its break-even; locking a profit across opposite sides (rarely clean between a pick'em platform and a sportsbook).
  • DTD / GTD / IR / Questionable / Out / Doubtful — injury designations. Depth chart order — rank at the position (1 = starter).
  • rowId, projectionId, lineKey — this Actor's unique row key; the over side's id shared by both sides of a line; the market key that pairs the sides.

🤖 AI Agents, MCP & RAG Integration

Sleeper Picks Scraper works natively as an on-demand data tool for AI assistants and autonomous workflows.

Connect this Actor directly to Apify AI, Claude Desktop, Cursor, VS Code, Windsurf, LangChain, or LlamaIndex via the official Apify MCP Server — no custom wrapper code needed:

{
"mcpServers": {
"apify": {
"type": "http",
"url": "https://mcp.apify.com?tools=logiover/sleeper-picks-scraper"
}
}
}
  • Schema-typed inputs & outputs: Clean JSON records with explicit .actor/input_schema.json and .actor/output_schema.json definitions so LLM agents infer parameters and parse fields accurately without HTML noise.
  • Bounded & predictable runs: Built-in volume caps and pay-per-result budget guards prevent runaway agent tool loops.
  • RAG & automation ready: Stream run datasets straight into vector stores, n8n, Make, Zapier, or webhook pipelines.

What the Actor does. It reads public data Sleeper serves to every visitor and app user: the open board, the promo board, the scoreboard and the player catalogue. It does not log in, use any account, bypass a paywall or access control, or collect personal data about Sleeper's users (pick counts are aggregates), and it does not place, change or cancel picks. It makes a few dozen requests per run.

What you are responsible for. Terms of service, gambling and data laws differ by country and state; Sleeper Picks itself is not available everywhere. Whether you may use this data, for what, and how you store or redistribute it is your decision and responsibility; this is not legal advice. If your pipeline handles personal data, GDPR, CCPA and similar rules apply to you.

Use it for analysis. Research, modelling, journalism, comparison, education, monitoring. The Actor is not a tip service: nothing here predicts outcomes, no strategy built on the data is guaranteed to win, and past games in recentGames are not forecasts. Never stake money you cannot afford to lose. If gambling stops being fun, contact your country's problem-gambling helpline and use the limits and self-exclusion tools your platform offers. The Actor is independent, carries no affiliate links and promotes no operator.

Changelog

2026-11-14. Verified live production health run (2026-11-14): extracted 35 structured records in 3.8s with valid schema output. Hardened internal proxy stall fallback and pay-per-event dataset instance budget protection.

2026-11-11. Added reliable terminal run status telemetry and structured Apify AI / MCP Server (mcp.apify.com) tool configuration. Added direct Issues & Reviews support links and refreshed live maintenance verification.

2026-11-01. Monthly health check (2026-11-01): a live run on the current build completed successfully and returned data. Input options, output fields and pricing are unchanged.

1.0 — 2026-10-29. First release. The live Sleeper Picks board with one row per line — the over and the under side by side, like a one-row-per-prop table — with multipliers, American odds, implied and no-vig probabilities, pick counts and popularity, recent form and hit rates, injury and profile columns, matchup, venue, TV channel and start time, and promo flags. rowShape: side writes one row per side instead (choosing a side filter switches to it), and includeRecentGames adds one row per past game. Filters for league, stat, player, team, side, multiplier, pick count and start window; soccer competitions, split-year seasons and promo twins handled; pay per result with a maximum-cost guard; direct connection first with a datacenter and a capped residential rescue; a SUMMARY record with the outcome and counts.

💬 Support, Issues & Reviews

A missing field, a stat label you would like, a league that behaves oddly: open an issue on the Actor page or send a message — we reply within 24 hours. Include the run id and, if you can, the input; the SUMMARY record in the run's key-value store usually tells us the rest.

  • Bug reports & target-site changes: Hit an unexpected empty result or selector change? Open a ticket in the Issues tab with your Run ID or sample input — we actively monitor issues and ship fixes fast.
  • Feature & field requests: Need an additional data attribute or filter? Request it in the Issues tab.
  • Enjoying this Actor? If it saved you engineering time, please consider leaving an honest review on Apify Store or bookmarking logiover/sleeper-picks-scraper — organic developer feedback helps others discover reliable tools and keeps maintenance prioritized.