Odds API pricing: per-request credits vs flat rent, with the maths

Most sports data vendors publish a price per request, per credit or per data point. That looks cheap on the pricing page and expensive the first time you poll twenty in-play cricket markets every second. This post works the numbers step by step, puts the pricing models side by side in rupees, and shows where each one hurts, including where our own flat rent is the wrong choice.

Three ways vendors charge, plus a fourth

Almost every odds API falls into one of four billing models. What matters is the unit you are charged for, because it decides whether the bill scales with traffic, products or revenue.

• Credits per request: each call burns credits from a monthly allowance. On The Odds API a call to the odds endpoint returns one sport and costs one credit per market per region requested, with tiers from a free 500 credits to 15 million a month.
• Data points: you pay per row of odds returned, not per call. TheRundown bills this way, and its cheaper tiers are delayed: 60 seconds on the $49 Starter plan, with real-time data only from $399.
• Flat monthly rent: a fixed rupee or dollar amount per feed, unlimited requests. This is how we sell match odds, fancy, bookmaker and score feeds.
• Revenue share: the vendor takes a percentage of gross gaming revenue. It is how our cricket exchange is sold, with the odds and fancy feeds included in the share.

What in-play cricket polling actually consumes

Take an evening with two live cricket matches. Each shows match odds, a bookmaker market and around eight open fancy (session) lines, so roughly ten markets per match and twenty in total. Most operators poll in-play markets every second, because a fancy line that is three seconds stale is a line a sharp player will hit.

• Step 1: 20 open in-play markets, each polled once per second, is 20 requests per second.
• Step 2: 20 requests per second multiplied by 3,600 seconds is 72,000 requests per hour.
• Step 3: multiplied by 24 hours is 1,728,000 requests per day, or about 1.7 million.
• Step 4: multiplied by 30 days is 51,840,000 requests per month, call it 52 million.
• Step 5, the generous case: polling only a 12-hour window is still 864,000 requests per day and about 26 million a month.

Five live matches with full fancy coverage means fifty to eighty open markets, and the same arithmetic gives 130 to 210 million requests a month.

Vendors priced side by side (USD to INR)

The rows below use published list prices as of September 2026, converted at a round ₹83 per US dollar, which is an assumption, not a quote.

• The Odds API, Starter: $30 per month for 20,000 credits, about ₹2,490. At the 20-market polling pattern that allowance lasts under 17 minutes.
• The Odds API, largest listed tier: $249 per month for 15 million credits, about ₹20,667. The 52 million request month needs roughly three and a half of these; even the 12-hour polling day exceeds it by about 11 million credits.

• TheRundown, Starter: $49 per month, about ₹4,067, for 25 million data points with a 60-second delay. Real-time data starts at $399 per month, about ₹33,117. A data point is a price row, so one poll of twenty markets with several runners each consumes many points.
• TheRundown, Max: $2,499 per month, about ₹207,417. Overage runs from $12.25 down to $1.25 per million data points, which is where high-frequency polling ends up.

• 7 Unicorn Gaming, Cricket, Tennis and Soccer Match Odds: ₹50,000 per month, about $602, unlimited requests, three odds sources, horse racing and baseball on the same plan.
• 7 Unicorn Gaming, Match Fancy: ₹50,000 per month, unlimited requests, covering Betfair, Khhadda market, manual and Type D fancy. Neither credit vendor above lists cricket session markets at all.
• 7 Unicorn Gaming, Bookmaker Match Odds: ₹40,000 per month; Match Odds and Fancy Combo: ₹90,000; Fancy, Match Odds and Bookmaker together: ₹130,000. All unlimited requests.

At real in-play polling rates the credit plans land in the same rupee band as flat rent and pass it at exchange scale, and neither carries the bookmaker and fancy markets an Indian cricket page needs.

A cricket latency budget: from ball bowled to suspended market

When a wicket falls, the source suspends the match odds and every open fancy line within moments, and you need that suspension on the player's screen before the player sees the wicket on a stream that is itself a few seconds behind. SharpAPI's guide on real-time odds makes the general point: freshness is bound by how fast the source captures the move, and with a REST feed your poll interval is added on top.

Our feed updates within a second of the source and passes suspension through as a status change on the same market object, which is why the live demo shows prices refreshing every second or so. A rough budget on a one-second poll:

• Source suspends the market on the ball: the part you cannot control, so compare providers on the live demo, not the brochure.
• Feed publishes the new status: under one second on our side.
• Your poller picks it up: on average half your poll interval, so 500 ms at one second and 5 seconds at a ten-second interval.
• Your bet-acceptance logic locks the market: a database write and a cache flush, tens of milliseconds.

The biggest lever you control is the poll interval. The arithmetic above pushes a credit-plan user towards ten-second polls, and every second added is a second of open bets on a dead price. For the failure case, we supply match odds from three sources and you switch from your side when one fails, so a source outage costs a config change rather than an evening.

Where per-request pricing punishes you

Credit billing suits a content site that refreshes odds every few minutes for a handful of matches. It punishes anything that looks like a real betting front end.

• In-play polling at one second, as calculated above, burns through a mid tier in hours.
• Fancy and session coverage, because every open line is another market to fetch.
• Multiple sports at once: cricket, soccer and tennis in the same evening triples the request count on a per-sport credit model.
• Reliability: a retry, a health check or a second data centre polling the same market doubles your spend, so the cheapest engineering decision is also the least resilient one.

Where flat rent punishes you

₹50,000 per month is the same bill whether you serve ten players or ten thousand, and whether there is a Test match on or nothing at all. Rent is the wrong choice for:

• Pre-launch and low-volume sites: while you are still building, a free or $30 credit tier is enough to develop against.
• Single-sport, pre-match-only products: a tipster site refreshing 30 soccer fixtures once an hour makes about 720 requests a day, and a credit plan will always be cheaper for that.
• Fixed cost before revenue: a new exchange pays full rent from month one, which is why our cricket exchange is sold on a share basis with feeds included, and the rent-based feeds are for operators who already have volume.

The crossover with revenue share is rent divided by the share percentage. At an assumed 15 per cent of gross gaming revenue, a ₹50,000 feed and a share deal cost the same at ₹50,000 ÷ 0.15 = ₹3,33,333 of monthly GGR, and the ₹130,000 stack at ₹8,66,667. Below those figures share is cheaper; above them rent is. The 15 per cent is an assumption for the example, not a vendor quote, and the full 12-month working is in our betting website cost post.

Our plans and what unlimited means

Every 7 Unicorn Gaming feed is a fixed monthly rent with unlimited requests: Cricket, Tennis and Soccer Match Odds ₹50,000 per month; Match Fancy ₹50,000; Bookmaker Match Odds ₹40,000; Match Odds and Fancy Combo ₹90,000; Fancy, Match Odds and Bookmaker Match Odds together ₹130,000. The Score API is ₹15,000 per month on the slow plan and ₹30,000 on the fast plan, and it is the feed fancy markets settle against.

Unlimited means what it says. Poll every open market every second, run a second poller in another region, add a health check on the event list once a minute, and the invoice does not change. Pricing is the same whether you take all three main sports or cricket only. The live prices are free on the Check API pages; when we checked on 18 September 2026 the demo event list showed 27 cricket events, each with match odds, bookmaker and fancy links.

Our view: we do not sell per request because in-play cricket makes per-request pricing a tax on doing it right. A one-second poll on every open fancy line is correct engineering for an exchange, and a model that charges more for the correct thing pushes operators towards the wrong one. The bill should be a number you put in a spreadsheet in January and forget.

Frequently asked questions

How much does a sports odds API cost?

From free developer tiers to a few hundred dollars a month on credit plans, and into the low thousands for real-time data-point plans. Flat-rent feeds for the Indian market sit at ₹40,000 to ₹130,000 per month depending on the markets you take.

Is there a free cricket odds API?

Free tiers offer a few hundred credits, or a capped and delayed data-point allowance, enough to build against but not to run a site. Our Check API pages show live cricket, soccer and tennis odds free with no key.

Do odds APIs charge per request?

Many do, as credits per call or data points per row. Some, including ours, charge a flat monthly rent with unlimited requests. Multiply the billing unit by your real polling pattern before comparing headline prices.

What counts as one request on a credit plan?

On The Odds API, one request for one sport, one market type and one region is one credit, so several markets or regions cost several credits. On TheRundown the unit is a data point, roughly one price row.

How fast should I poll in-play cricket odds?

Every second. The feed updates within a second of the source, longer intervals save nothing on a flat-rent plan, and each extra second is time a suspended market stays open on your site.

Key takeaways

• Twenty in-play markets polled once per second is 1,728,000 requests a day and about 52 million a month; run that against any credit or data-point tier before you sign.
• At real in-play polling rates, credit plans reach the same rupee band as flat rent, and they do not carry bookmaker or fancy markets.
• Flat rent loses when you are pre-launch, single-sport or pre-match-only; build on a free tier, move to rent when volume arrives.
• Your poll interval is the largest part of the latency budget you control, and per-request pricing pushes it the wrong way.