Underdog Fantasy Player Props avatar

Underdog Fantasy Player Props

Pricing

from $0.15 / 1,000 player props

Go to Apify Store
Underdog Fantasy Player Props

Underdog Fantasy Player Props

Collect Underdog pick’em player props with paired higher/lower sides, lines, native payout details and saved change tracking. Filter supported leagues, players, teams, stats and UTC game dates.

Pricing

from $0.15 / 1,000 player props

Rating

0.0

(0)

Developer

Philip Kirkbride

Philip Kirkbride

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

3 days ago

Last modified

Categories

Features

Limited permissions

Share

Collect Underdog Fantasy pick'em player lines for DFS research, with Higher and Lower selections together in each prop row. Get native player and game IDs, game times, stat labels, lines, side availability and payout details. Save a baseline to monitor changes across scheduled runs.

Use cases

  • Compare DFS lines across pick'em providers using consistent player, stat and game fields.
  • Compare a prop with sportsbook lines after checking the player, event, period and settlement rules.
  • Supply current line and availability observations to a model without mistaking fantasy multipliers for betting odds.
  • Build alerts for new props, line moves, payout changes, suspended selections and removals.

Also available

The player props suite: pick the job.

ActorUse when
Underdog Props (this Actor)Underdog higher/lower lines and payouts
PrizePicks PropsPrizePicks lines and projection changes
Sleeper Picks PropsSleeper lines, multipliers and changes
DraftKings Pick6Pick6 props, multipliers and alternates
BetRivers Odds & PropsBetRivers and Unibet lines plus player props
Sports Injury ReportsInjury status changes across leagues

Quick start

The Console prefill requests up to 20 NBA/NFL props:

{"leagues":["nba","nfl"],"includeAlternates":false,"includeLive":false,"maxResults":20,"onlyChangesSinceLastRun":false,"proxyConfiguration":{"useApifyProxy":false},"allowPartial":false}

Running {} announces “Sample run (...)”, returns at most 20 props across supported leagues and sets OUTPUT.sampleRun: true. During a dead period it succeeds with an honest empty summary. An explicitly empty filter, such as leagues: [], ends the run as no_results and names the field. Omit filters to include everything supported. Lists accept at most 200 distinct values of at most 200 characters each. Empty, whitespace-only, duplicate normalized values and values without Latin letters or digits are skipped into OUTPUT.skippedSearches; when no usable entry remains the run ends as no_results. Values within a list are combined; filters across fields must all match. Disjoint known sports and leagues, or nextDays with explicit dates, fail before collection.

Input

FieldMeaningOmitted behavior / limits
sportsSport names or supported league codesAll observed sports; basketball, american_football, canadian_football, ice_hockey, esports, tennis, golf, mma
leaguesNBA, NFL, NHL, NCAAF/CFB, NCAAB/CBB, CFL, CS/CS2, DOTA, LOL, tennis, PGA, MMA/UFCAll supported leagues; availability varies with the slate
statTypesExact native or normalized stat keysAll; examples: receiving_yds, receiving_yards, points
periodsNormalized output values: game, first_period, second_period, third_period, fourth_period, first_half; full game is gameAll periods; unknown values are skipped into OUTPUT.skippedSearches
playersExact full names or native player IDsAll; name comparison ignores accents and punctuation
teamsExact player-team abbreviation or full nameAll; compares with player.team and player.teamName
gameDateFromInclusive UTC game date, YYYY-MM-DDNo lower date bound
gameDateToInclusive UTC game date, YYYY-MM-DDNo upper date bound; must not precede gameDateFrom
nextDaysToday through today + N − 1 in UTCOmitted; integer 1–30; cannot combine with explicit dates
includeAlternatesInclude other offers present for a marketfalse; does not promise a complete alternate ladder
includeLiveRetain live and non-pregame offers presentfalse; no promise of complete in-play coverage
maxResultsMaximum delivered prop/change rows20 for bare input, otherwise 5,000; integer 1–10,000
onlyChangesSinceLastRunBaseline followed by changesfalse
stateKeyPersistent history name shared by matching filtersdefault; non-empty string, at most 200 characters
proxyConfigurationOptional connection settingsOmitted; no configured service required by default
downloadTimeoutSecsDeadline for acquiring the complete board90 seconds; integer 1–1,200; run timeout also covers saving
allowPartialDeliver complete, joined rows from an incomplete boardfalse; incomplete collection normally fails

To collect only full-game props, use:

{"periods": ["game"]}

Period values ignore case and surrounding whitespace. An explicit list excludes rows with an unknown (null) period. Period filtering happens before enrichment and delivery; excluded props incur no record fee. The run start fee still applies.

Output

One dataset row represents one player/stat market/provider, including both sides. A missing selection remains present with available: false. Native payout values remain under sides; fantasy multipliers are never converted to probabilities or sportsbook prices. Unknown normalized stats, ESPN IDs, settlement rules and overtime coverage remain null. matchKey is a comparison aid built from league, normalized player name and team (league and name for solo sports); it is not a verified cross-provider ID.

Rows also include recordKey, stable propKey, snapshotKey, canonical version 1.0, nullable quote version 1.1, source timestamps, collection timestamps, suspension flags and optional alternates. Changed rows add changeType, changedFields and previous. Compare the same period and scoring rules; a first-half line differs from a full-game line.

The key-value store OUTPUT contains row counts by sport, league and stat, request and byte totals, failures, missing-data diagnostics, sample/baseline flags, caps reached and summaryLine. The terminal status message matches that summary. Summaries and errors are not dataset rows.

Output example

Selected fields only.

A real row delivered by the Actor entrypoint from the complete November 11, 2026 capture:

{
"propKey": "underdog:0004b884-9d7b-412d-a40a-42310d628b3e:manual",
"source": "underdog",
"league": "ncaaf",
"player": {
"nativeId": "93398dff-c2c8-4f6c-8ee3-f28b2050ddbc",
"name": "Ty Clark",
"team": "WAKE",
"teamName": "Wake Forest Demon Deacons",
"teamNativeId": "629be65c-616e-4c16-9b22-6228a373b826",
"position": "RB",
"espnPlayerId": null,
"sportradarId": null
},
"statTypeNative": "rush_rec_tds",
"statType": "touchdowns",
"line": 0.5,
"espnPlayerId": null,
"matchKey": "ncaaf:tyclark:wake"
}

Pricing

$0.00015 per prop record ($0.15 per 1,000) plus $0.01 per run start. Under pay per event there is no separate compute or proxy charge.

Higher and Lower together count as one prop record; nested alternates do not add records. A 20-record sample costs $0.013 ($0.003 + $0.01). All 9,773 records in the captured board cost $1.47595 ($1.46595 + $0.01). All 3,112 captured NFL props cost $0.4768. An unchanged check delivers zero records and costs the $0.01 run start. A start-only charge limit does not advance the baseline. These examples assume the stated rates and sufficient charge and output limits; future board sizes vary.

Only delivered records count toward the record fee. A charge limit may deliver fewer rows; the summary reports “allowed N of M”, and undelivered changes remain pending for later runs. For example, a $0.01045 run limit permits the start plus three records at these rates. Set maxResults: 10000 to request the whole captured board; an explicit scope otherwise defaults to 5,000.

Scheduling and monitoring

Run this input repeatedly with the same stateKey and filters:

{"leagues":["nfl"],"statTypes":["receiving_yards"],"maxResults":20,"onlyChangesSinceLastRun":true,"stateKey":"nfl-receiving-watch"}

The first run emits a labelled baseline. Later runs emit new, line, payout, status or removed records, with prior values. A row can list several changed fields. Unchanged props produce no row. Only delivered observations advance saved history; capped rows can appear next time. A new state name starts fresh history. Changing filters, including adding, removing or changing periods, starts a separate baseline in a separate history; changing the cap preserves it.

Incomplete observations never infer removals. A completely empty league preserves its prior props conservatively. Check stateSaved and warnings; a storage failure can cause changes to repeat. Only one run may write the same history at a time. An overlapping run fails before collection and names stateKey; a different state name can run independently. If a run is killed, its history lock can remain until its timeout plus 60 seconds. History is bounded to 10,000 observations; new props remain pending if delivered removals must first free space. Corrupted or oversized history fails with a stateKey error instead of resetting it. Earlier version 1 history records remain readable.

Run time

A filtered sample still acquires the whole available board. A complete platform observation on November 11, 2026 received about 2.85 MB compressed / 25.997 MB decoded in 4.2 seconds, with about 105 MB peak memory during the observed download. The saved gzip is 2.636 MB. Small output limits reduce saved rows, but do not reduce acquisition. There is one board request per run.

The default download deadline is 90 seconds; the default run ceiling is 900 seconds at 512 MB. Complete-capture local replays at approximately one eighth CPU delivered 9,773 rows in 66.8 seconds using 308 MiB peak memory and a saved baseline in 79.9 seconds using 308 MiB. With every line deliberately moved, the change replay delivered all 9,773 changes in 87.8 seconds using 340 MiB; its unchanged repeat delivered zero rows in 33.7 seconds. The 20-row sample finished in 19.4 seconds using 89 MiB and normalized only the requested sample. Acquisition still reads the complete document. These are replay measurements; platform storage and network times can differ. Results are saved incrementally in small batches. Near the run deadline, collection stops with a partial summary and preserves only delivered history so the remainder can appear next time.

A cut-off collection fails with a reason. Explicit allowPartial: true retains fully read, joined props and reports missing arrays, unfinished records and unresolved lines. Unseen remainder size is unknown. Partial results never infer removals.

Limits and freshness

The complete capture contains NBA, NFL, NHL, college football (NCAAF/CFB), college basketball (NCAAB/CBB), CFL, Counter-Strike (CS/CS2), Dota 2, League of Legends, tennis, PGA-labelled golf and UFC-labelled MMA. These labels describe source coverage, not a guarantee of every event or future slate. Golf's captured event is the Austin Championship. There is no MLB or WNBA evidence in this capture. Solo sports use participant names or a round title and have no player team.

All 9,773 captured lines became rows, with zero skipped lines. This snapshot has balanced pregame offers only. hasAlternates is a source hint; alternate ladders, discounts, boosts, live lines and expirations have no real examples here. Suspended options and missing sides are observed and retained. Native payout metadata stays under sides.payoutInfo; multipliers are separate from American and decimal prices.

Data describes the board during collection, not historical or guaranteed closing lines. sourceUpdatedAt is the latest timestamp among a line and its options, or null. All 9,773 rows have name-based matchKey values, with no collisions across the 1,614 native players in this capture; this does not verify any cross-provider identity. Zero of 1,614 players can be mapped to ESPN IDs from the supplied document alone. ESPN player/event IDs remain null; no enrichment is performed. CFL has abbreviations but no full team names in the document.

Unknown stats retain statType: null; 7,173 captured rows have normalized stat aliases. Native stat labels preserve source scope. The 290 offers scoped to map/game subsets (including ten three-map fantasy offers) have period: null and quote: null because quote version 1.1 has no corresponding period. They remain honest native prop rows with their scope in statTypeNative and the canonical native stat; they are not labelled full-game quotes. The other 9,483 rows carry quote version 1.1. Scoring rules, overtime coverage and entry-size payouts are unverified.

Collection is bounded at 128 MiB received and 256 MiB decoded. Output is capped at 10,000 rows per run. A sample does not promise equal representation by league. All date filters use UTC. Filter values compare exactly after case, accent, Latin-letter and punctuation normalization; substrings do not match. An unknown value succeeds empty and is named in OUTPUT.unmatchedFilters and the terminal summary.

FAQ

Can I get only one player or stat? Yes. Use an exact name/native ID or native/normalized stat key. Filters apply after the board is acquired.

Are both selections always available? No. One-sided and suspended props retain explicit unavailable selections. Do not infer a missing price.

Does it provide historical lines? No. Schedule repeated runs to build observations from your own collection times.

How do I export results? Use the default dataset's JSON, CSV, Excel or other supported Apify exports. Keep nested sides in JSON for lossless payout detail.

What if nothing matches? A complete valid collection succeeds empty and explains this in OUTPUT. A failed collection ends FAILED; an explicitly permitted partial result names its gaps.