Darts Player Form: Averages, Checkout % & 180s

Check a darts player's form over any date window from the public statistics on Darts Orakel. For each player ID you give, the actor reads the player's match list three times (averages, 180s and checkout percentage), checks that the three lists describe exactly the same matches, and returns one row per player and window: the weighted three-dart average, checkout percentage, 180 count, wins and losses, plus every match as nested evidence and an explicit statement of what the source covered.

It is built for people who refresh a fixed watchlist before updating their own models or notes. A quiet window is still a useful, validated answer (no_matches), and is kept clearly apart from a failed fetch.

Input

{
    "playerIds": ["2367"],
    "dateFrom": "2026-01-01",
    "dateTo": "2026-10-16",
    "maxRows": 20,
    "watchName": null,
    "requestIntervalMs": 2100
}

• playerIds: 1 to 20 numeric Darts Orakel player IDs, from a public profile URL such as dartsorakel.com/player/details/2367/maikel-verberk. Duplicates are checked once. Names, URLs and other text are rejected; the actor does not search for players.
• dateFrom / dateTo: inclusive source calendar dates (YYYY-MM-DD), at most 366 days, not after today's UTC date. These are the source's own match dates, not UTC instants. Compute rolling windows in your scheduler.
• maxRows: stop after this many delivered checks (1 to 20).
• watchName (optional): compare each check with the last check of the same player and window under this name. Rows then say new, unchanged or changed. Leave it empty for no stored state (untracked).
• requestIntervalMs: pause after each source response, 2,100 to 60,000 ms. One request at a time.

Output

One row per delivered check. Main fields:

Field	Meaning
playerId, playerName	The requested ID and the player name as the source publishes it (null when no record was returned).
checkStatus	form_available, or no_matches for a validated window with no eligible match.
emptyReason	empty_source_window (the source returned no records) or only_byes; null otherwise.
matchCount, wins, losses	Completed non-bye matches in the window.
threeDartAverage	3 × total points ÷ total darts over the window, six decimals. null if any match lacks the figure.
checkoutPercent	100 × total successful doubles ÷ total attempts, six decimals. null if any match lacks the figure.
count180	Sum of reported 180s. null if any match lacks the figure.
statTotals	The exact integer totals behind those rates, including the known part of a partial metric.
coverage, missingStatFlags	How many matches carried each statistic, byes excluded, and which metrics were partial.
matches	Every included match: date, event, opponent, result, scores, per-match figures and the source's raw parts.
checkEvidence	Per view: counts the source reported, rows returned, a SHA-256 of the raw body and the time it was checked.
contentHash, changeType, previousContentHash	A hash of the check's content and, with a watch name, how it compares to the previous check.

Example (shortened, from a real run of the default input):

{
    "playerId": "2367",
    "playerName": "Maikel Verberk",
    "checkStatus": "form_available",
    "matchCount": 84,
    "wins": 55,
    "losses": 29,
    "threeDartAverage": 87.109247,
    "checkoutPercent": 36.219641,
    "count180": 155,
    "statTotals": {
        "averageNumerator": 273494,
        "averageDenominator": 9419,
        "checkoutNumerator": 343,
        "checkoutDenominator": 947,
        "count180KnownTotal": 155
    },
    "missingStatFlags": ["unknown_score_unit"],
    "changeType": "untracked"
}

Use JSON for the full nested row. For a spreadsheet, export CSV/Excel with the platform's fields option on the scalar columns (for example playerId,playerName,checkStatus,matchCount,wins,losses,threeDartAverage,checkoutPercent,count180,changeType,fetchedAt). Exporting nested matches as extra lines does not create extra billed checks.

A free SUMMARY record in the run's key-value store lists every player's outcome, every error with its code, the number of source requests and bytes, and how many rows were stored and charged.

What the figures mean, and what they do not

• Completeness is the source's, not the real world's. A check covers exactly what Darts Orakel returned for the window; all three lists must report complete, matching match sets or nothing is delivered. realWorldCompletenessGuaranteed is always false.
• Points/darts and doubles/attempts are inferred from the site's labels and from the fact that every sampled per-match figure equals 3 × stat1 ÷ stat2 (average) or 100 × stat1 ÷ stat2 (checkout). The actor re-checks that agreement on every match and rejects a view that disagrees.
• Scores have no unit. A score such as 6-5 may be legs or sets depending on the format; scoreUnit is always unknown.
• Missing figures are never zeros. A match without a figure lowers coverage; reported zeros stay zero.
• An empty window does not prove anything about the player. It does not show that the ID exists, that the player did not play, or that the source is complete (playerExistence is unknown).
• Byes are excluded from form and counted in coverage.excludedByeCount.
• No betting advice, odds, predictions or premium statistics.

Pricing

Pay per event:

• $0.00005 per run start, charged even when the run ends in an error.
• $0.003 per delivered check (one dataset row), including validated empty and unchanged checks.

Errors, skipped players and the SUMMARY are never charged as checks. Ten players in one run cost $0.03005; twenty quiet (empty) windows cost $0.06005. A spending limit below $0.00305 cannot cover one check, so the run stops before contacting the source.

Limits

• 20 players per run, three requests per player plus one robots.txt request, at most 122 requests in total.
• One request at a time with at least 2.1 s after each response, 20 s per request, one retry for timeouts, rate limits and server errors. Responses over 4 MiB, more than 16 MiB per run, or more than 1,000 matches per view stop the run.
• Work stops 810 s after the run starts; the remaining players are reported as not attempted. At slow pacing (for example 60 s) only a few players finish.
• Rows up to 1 MiB each and 4 MiB per run.
• Storage traffic (rows, read-backs, SUMMARY and watch state, counted as sent plus 1 KiB per call) is capped at 16 MiB per run. The run stops before passing it and always keeps room for the final SUMMARY.
• A blocked or challenged request (HTTP 401 or 403, a login page or redirect, bot check) stops the whole run: the actor never retries with other addresses, headers, browsers or proxies.
• Runs at 256 MB; a run started with more memory is rejected before any request.
• Watch state keeps only hashes and dates: at most 500 player-windows and 128 KiB per watch name, each kept for 30 days after its last check, then treated as new. Do not run two runs with the same watch name at the same time. A run that cannot release its watch name or save its final SUMMARY ends as failed, even after delivering rows.

Source and your responsibilities

Data comes from the public, login-free pages and JSON of dartsorakel.com. Its terms reserve rights and restrict redistribution; this actor is not affiliated with or endorsed by Darts Orakel and is not a licence to reuse its data. Rows contain player names and sporting results as published. You are the controller of any personal data you collect and are responsible for your use of it. Accuracy for modelling or betting is not guaranteed.