Migrating from the-odds-api in 5 minutes

The whole migration: one URL change

If you're using the-odds-api.com directly, swap the base URL to parlay-api.com. That's it.

# Python (requests)
- BASE_URL = "https://api.the-odds-api.com/v4"
+ BASE_URL = "https://parlay-api.com/v1"

r = requests.get(
    f"{BASE_URL}/sports/baseball_mlb/odds",
    params={"markets": "h2h", "regions": "us"},
    headers={"X-API-KEY": "YOUR_PARLAYAPI_KEY"},
)

If you're using their official the-odds-api SDK, point it at our base URL via the SDK's config option, or migrate to plain requests / fetch. Most users do the latter; the response is identical.

// JavaScript (fetch)
- const BASE = "https://api.the-odds-api.com/v4";
+ const BASE = "https://parlay-api.com/v1";

const events = await fetch(
  `${BASE}/sports/baseball_mlb/odds?markets=h2h&regions=us`,
  { headers: { "X-API-KEY": YOUR_KEY } }
).then(r => r.json());

Endpoint compatibility

Sport keys: 100% identical for US majors

You don't need to change a single sport_key string for these:

• baseball_mlb, basketball_nba, americanfootball_nfl, icehockey_nhl
• americanfootball_ncaaf, basketball_ncaab, basketball_wnba
• mma_mixed_martial_arts, mma_ufc
• soccer_epl, soccer_spain_la_liga, soccer_germany_bundesliga, soccer_italy_serie_a, soccer_france_ligue_one
• soccer_uefa_champs_league, soccer_uefa_europa_league, soccer_conmebol_copa_libertadores

We add 60+ sport keys the-odds-api doesn't carry: French Ligue 2, Brazilian Serie B, Korean KBO baseball, Japanese NPB, Polish Ekstraklasa, Swiss Super League, Czech table tennis, esports leagues (CS2, LoL, Valorant, Dota 2), and more. Full list at /v1/sports?all=true.

Bookmaker keys

The authoritative list is GET /v1/bookmakers. Below is what's live today. If the key isn't in /v1/bookmakers, it isn't ingested; we do not maintain placeholder entries.

Shared with the-odds-api (same key string)

draftkings, fanduel, betmgm, caesars, betrivers, bovada, pinnacle, fanatics, bet365.

Some books the-odds-api lists (PointsBet US, William Hill US, Wynn, Unibet US) are not in our ingest. We avoid maintaining a dead key just for parity; if you depended on those, check the corresponding feed below or contact support.

Additional books we carry that the-odds-api does not

• Sharp anchor: pinnacle (US-blocked at the-odds-api)
• Exchange: novig
• Prediction markets: kalshi, polymarket, robinhood
• DFS pick'em: prizepicks, underdog, sleeper, betr
• US bargain / regional: fliff, hardrock, parx
• European: bwin, unibet, pmu, betclic, winamax

Pricing comparison

The honest per-credit math. Read the rightmost column. Bigger plans on the-odds-api are cheaper per credit than our equivalent tier; smaller plans (Free, Starter, Pro) are dramatically cheaper on ParlayAPI. Pick the row that matches your monthly volume.

Credit costs per call vary by endpoint (1 credit for most reads, 5 credits for /v1/sports/{key}/compare and /v1/clv). Hit /v1/usage on your account to see your actual burn rate before migrating.

Hit your credit limit mid-Sunday?

Every September the same thing happens: an app that ran fine all summer dies during the 1 o'clock NFL window with a quota error. The cause is credit math, not traffic: a credit is not a request. Per the-odds-api's own usage docs (read 2026-08-25), one call to their /odds endpoint costs [markets specified] × [regions specified] credits, and player props are only reachable through per-event calls billed on the markets each event returns.

The worked example

Polling markets=h2h,spreads,totals&regions=us is 3 × 1 = 3 credits per poll, not 1. Add player props on a typical 13-game Sunday main slate — 13 per-event calls, each returning ~5 prop markets in one region — and each full sweep of the slate costs ~65 credits.

# One NFL Sunday, ~1:00 pm to ~11:30 pm ET (630 minutes), on the-odds-api
Game lines every 60s:    630 polls  x 3 credits   =  1,890 credits
Props every 2 minutes:   315 sweeps x ~65 credits = ~20,475 credits
                                            total   ~22,365 credits
# Their 20,000-credit month is gone before Sunday Night Football ends.

Their 20K plan size per their pricing page (read 2026-08-25). In fairness: their per-event billing counts only markets actually returned, so a thin slate costs less than this worst case.

The same Sunday here

What those credit allowances cost is on /pricing — and note the game-line formula is identical here, so if props aren't your problem, migrating changes what a credit costs you, not how many each poll burns. Historical odds are 10 × markets × regions on both APIs. Every paid response here carries an x-requests-last header reporting the exact credit charge for that call, and streaming endpoints (/v1/ws, /v1/sse) can replace polling outright on tiers that include them — see /pricing.

The 60-second migration checklist

1. Sign up — key issued in ~30 seconds, no card, 1,000 free credits to validate with.
2. Swap the base URL: https://api.the-odds-api.com/v4 → https://parlay-api.com/v1. Same X-API-KEY header, your new key.
3. Keep your sport keys and markets/regions params — identical for US majors (americanfootball_nfl et al.).
4. Delete the per-event props loop; replace it with one GET /v1/sports/americanfootball_nfl/props (flat 3 credits).
5. Watch x-requests-last on each response and /v1/usage for the running total — no surprise math.
6. Run both feeds through one slate, diff the responses, then cut over.

What's actually different (the honest part)

Things you gain

• Player props on the dedicated /v1/sports/{key}/props endpoint, with full market-key support (player_points, player_anytime_td, etc.)
• WebSocket streaming at /v1/ws/odds/{key} and Server-Sent Events at /v1/sse/odds/{key} for push updates instead of polling
• Prediction-market integration (Kalshi, Polymarket, Robinhood Event Contracts) on supported events
• DFS pick'em ingestion (PrizePicks, Underdog, Sleeper, Betr)
• Side-by-side line comparison at /v1/sports/{key}/compare (also aliased to /v1/sports/{key}/best-line) for line-shopping bots
• Pinnacle prices, which are blocked from the-odds-api in the US market
• Free tier with 1,000 credits/month (2x the-odds-api's free tier of 500)

Things to know

• Player prop closing-line history: 30M+ prop closing lines back to June 2022, dense from spring 2026 forward. Game-line closing history is continuous from 2005 for NFL, 2017 for NBA.
• Refresh cadence is per book, not uniform. Hot-path books (Pinnacle, FanDuel) commit at 2 to 30 seconds typical. US majors (DraftKings, Caesars, BetMGM, BetRivers, Fanatics, bet365) commit at 30 to 120 seconds depending on cadence. European books (PMU, Unibet, Bwin, Betclic, Winamax) at 60 to 180 seconds. Per-source live numbers are at GET /v1/status and GET /v1/meta/provider-state; treat the documented values as steady-state guidance, not strict SLA. Business+ adds WebSocket and SSE streaming for push updates with sub-second wire latency from the broadcast layer (still bounded by upstream cadence).
• API response latency p50, measured 2026-05-15 post infrastructure migration: /healthz ~500 ms, /v1/sports/{key}/odds 600 to 900 ms, /v1/sports/{key}/props 800 to 1200 ms (limit 5000), /v1/sports/{key}/compare 1.5 to 2.5 s, /v1/historical/coverage ~250 ms warm.
• Large-result endpoints emit pagination headers (x-result-page-size, x-result-limit, x-result-offset, x-result-has-more, x-next-offset) so you can walk pages without guessing. Body shape stays a flat list, preserving back-compat.
• Some niche the-odds-api sports (cricket regional leagues, certain Asian markets) we do not cover yet. Check /v1/sports?all=true first.

Common gotchas

Authentication header

Both APIs accept X-API-KEY as a header. We also accept ?apiKey= as a query param for compatibility with code that uses the-odds-api's older docs. Either works.

Rate limits

None on paid tiers. Burst at whatever rate your client wants; you're metered on monthly credits, not per-second throughput. Free trial keeps a 60 req/s cap purely for anti-abuse during evaluation. If you need >10,000 req/s sustained from a single key (almost always a misconfigured retry loop), contact support to shard across keys.

Response timestamps

Same commence_time ISO-8601 UTC format. Same last_update per bookmaker on the live odds endpoints. No code change needed there.

One exception, on /v1/historical/sports/{key}/odds only: last_update and commence_time are nullable, and are null on most rows today. We serve the archive row's real capture time where we have one and null where we do not, rather than a placeholder; last_update_source tells you which (archive_snapshot, unavailable, or error_no_capture_column). Measured 2026-08-14: 68 of 397 bookmaker blocks stamped on a representative MLB date, and none on blocks carrying a moneyline. Guard both fields before parsing them on the historical route.

Bookmaker filter

Same ?bookmakers= comma-separated key list. We support all keys the-odds-api does plus the additional ones above.

Ready?

Sign up gets you an API key in 30 seconds. No credit card. Free tier supports real testing volume (1,000 credits/month).