Frequently asked questions

Real-time sportsbook odds API. Aggregates moneyline, spread, total, alt-line, and player-prop pricing from every major US-licensed sportsbook plus exchange and prediction-market books. Updates within roughly 1 to 3 seconds of upstream changes. REST + WebSocket on a single API key.

Three differences: latency (sub-second on WebSocket vs polling-only on most peers), data honesty (every price comes from a real bookmaker endpoint, no derivation or approximation, see how sportsbook APIs work), and built-in tooling (free /arbs, /ev, /middles, and /board visualizers backed by rate-limited try endpoints, plus reference clients in seven languages).

Every major US-licensed book (DraftKings, FanDuel, BetMGM, Caesars, ESPN BET, BetRivers, Fanatics, Hard Rock, WynnBET, PointsBet, SuperBook, and more), sharp anchors (Pinnacle, Circa, BetUS, Bookmaker), exchanges (Sporttrade, Novig, ProphetX), and prediction markets (Kalshi, Polymarket). Live coverage matrix at /coverage.

REST cached snapshot is 1 to 3 seconds behind upstream; WebSocket pushes within 200 to 600 ms for most US books. Per-book observed latency tracked at /v1/meta/source-quality and visualized at /uptime.

No. /v1/status, /v1/meta/source-quality, /v1/meta/parser-coverage, /v1/try/{sport}/odds, /v1/try/{sport}/arbitrage, /v1/try/{sport}/ev, /v1/try/{sport}/middles, /collections/postman.json, /badge/{kind}.svg, and all HTML pages are no-auth, rate-limited to 60 requests/hour per IP. The full API requires a free API key from /signup.

Free tier has a 1,000 request/month allowance, no credit card. Paid tiers: Starter $5/mo (20k requests), Pro $20/mo (100k), Business $40/mo (1M plus WebSocket), Enterprise $100/mo (5M), Scale $200/mo (50M plus sub-1s WebSocket). Full breakdown at /pricing. Project your monthly bill at /cost.

Free tier: 60 req/h per IP on try endpoints, 1,000 req/month on authenticated endpoints, 60 req/sec anti-abuse burst cap. Paid tiers have no per-second rate limit; you burst at whatever rate your client wants and are metered on monthly credits. WebSocket has no request cap but caps concurrent subscriptions per tier. We return 429 with Retry-After only if a single key blows past the 10,000-rps app-safety ceiling (almost always a misconfigured retry loop). See /docs/best-practices.

Yes. WebSocket on scale tier and above (wss://parlay-api.com/v1/ws?key=...). SSE fallback at /v1/sse/odds. Full docs at /docs/websocket.

A bet where your estimated win probability exceeds the implied probability of the price you're getting. Public +EV scanners use a sharp book as fair-value anchor: devig the sharp's two-sided price into no-vig probabilities, compare every other book's offered price against that fair probability. Our /v1/try/{sport}/ev endpoint does exactly this; /ev visualizes it live.

When opposing sides of the same market are priced such that the sum of their implied probabilities is less than 1.0, you can bet both sides at different books and lock a guaranteed profit. Our /v1/try/{sport}/arbitrage surfaces moneyline arbs; /arbs visualizes them live; the full endpoint covers spreads, totals, and player props.

On totals: bet Over X.5 at one book and Under Y.5 at another book where X < Y. If the final total lands in (X, Y), both bets win. On spreads: bet Team A -X.5 at book A and Team A +Y.5 at book B where X < Y. If Team A wins by a margin in (X, Y), you middle. Our /middles page surfaces totals middles live.

Closing-line value computed against the sharp book's no-vig fair probability rather than the vigged closing price. Why it matters: comparing vigged sharp vs vigged bet double-counts the vig and overstates edge by the overround. Pro graders use no-vig CLV. Full math at /blog/no-vig-clv-explained.

Direct from each book's public-facing endpoint. We don't derive prices and we don't approximate. If a book doesn't publish a particular market, we don't fabricate it. Per-book observed update latency at /v1/meta/source-quality.

Yes, on growth tier and above. MLB, NBA, NFL, NHL, EPL, MMA player props from every book that publishes them. Use markets=player_props on /v1/sports/{sport}/odds or subscribe via WebSocket.

Yes. Add live=true on /v1/sports/{sport}/odds for in-play markets only, or use /v1/inplay/* for the dedicated surfaces. Live arbs at /v1/inplay/arbs.

Yes. /v1/historical/closing-lines.csv exports closing-line snapshots. /v1/historical/odds returns timestamped per-market history. CSV exports cached 6h, free for all signed-in users.

Yes, production-grade clients in seven languages: Python, JavaScript (Node + browser), Go, Ruby, Java, PHP, and C#. Single-file, stdlib-only or minimal-deps, with retries, Retry-After honoring, idempotency keys, typed exception envelope. Browse at github.com/JacobiusMakes/ParlayAPI/tree/main/examples.

Yes, at /openapi.json. Browseable docs at /docs and /redoc. Postman 2.1 collection auto-generated from the spec at /collections/postman.json.

Yes, HMAC-signed alert webhooks for source-quality breaches and edge-detection triggers. Signature spec in /docs/best-practices; Slack/Discord relay reference at examples/source_quality_alert_relay.py.

Open an issue at github.com/JacobiusMakes/ParlayAPI/issues or email support with the X-Request-ID from your response. Source-quality SLA tracked at /v1/meta/source-quality.

99.95% on scale tier with a documented credit policy. Free and starter tiers are best-effort. Live status at /status, historical incident log at /status/history. Failover architecture documented at /blog/sub-1s-freshness.

Yes. ParlayAPI is in production at sports-data startups, model-running shops, and consumer-betting agents. Integration guidance at /docs/best-practices: auth, retries, idempotency, WebSocket reconnect, degraded-mode handling, observability, security, cost.