# How We Test and Verify: Our Methodology Source: https://cricketin.org/how-we-test [By Rajeev Sathe](/authors/editor-in-chief), Editor-in-Chief. [Reviewed by Karthik Iyengar](/authors/data-editor), Data Editor. Updated 31 Aug 2026. Editorial policy: https://cricketin.org/editorial-policy 18+ Betting money on cricket carries a direct risk of financial loss, and it can become an addiction. Real-money online gaming is banned across India under the PROG Act 2025, which came into force on 1 May 2026. If betting is already costing you money you need, or sleep, or people close to you, the free Tele-MANAS helpline is **14416**, and it runs in 20 languages. Our [page on help for gambling harm](/guides/help-for-gambling-harm-india) lists what else exists. --- Read Cricketin’s method for checking sources, logging observations, calculating overround, and showing the real cost of wagering rules. Cricketin checks each claim against a source. We note the date and the input. We show the steps behind our maths. Examples here teach a method, not a rating. This page sets out our log, our overround sum, and the way we work out the real value left after a wagering rule. ## This method explains a claim without selling a service We use one method for source checks and number checks. A reader should be able to take the same inputs and reach the same answer. We do not use it to rank operators. We give no scores, no badges and no verdicts, and we publish no list of places to play. The aim is to explain risk and terms, not to steer a reader into a money game. Section 6 of the Promotion and Regulation of Online Gaming Act, 2025 bars direct and indirect ads for online money games. It also bars any nudge to play ([the official Act text](https://upload.indiacode.nic.in/view-casepdf?type=act&id=AC_CEN_45_0_00004_2025-32_1778133559220)). The law took effect on 1 May 2026 ([Nishith Desai analysis](https://nishithdesai.com/research-and-articles/hotline/gaming-law-wrap/setting-the-rules-of-the-game-indias-online-gaming-law-comes-into-force-15586)). That is why our method stops at teaching and never becomes a ranking. ## What we check ourselves We check four things ourselves. We read the source named next to a legal or tax claim. We work out numbers from stated inputs. We test links and wording for errors. We log what we saw, and when. A sum is only as good as its inputs. We never call a general example a live market reading. We label a made-up case as made up. We call a number an operator measure only when we hold the dated record needed to redo it. On data and safety topics we also check the scope. A guide about [data privacy on gaming apps](/guides/data-privacy-on-gaming-apps) is not a claim about how we handle data. Our own habits appear only in [the privacy rule](/privacy-policy), once the owner has checked them. ## Every check needs a dated log For each check that matters we record the page URL and the subject. We add the source URL, the date and time in India, and the exact input. We name the person or role who ran it, the result and any limit. **Table 1. Fields in a Cricketin check record** | Field | What we record | Why it matters | |---|---|---| | Page and claim | The page URL and the exact statement checked | It shows what the record supports | | Source or input | Source URL, quoted section, or numbers used | Another person can inspect the basis | | Observed date | Date and time in India | Facts and page settings can change | | Method | Reading, calculation, link test, or rule review | It separates a fact check from a calculation | | Result | The result and unit, such as percent or rupees | It prevents a vague result | | Limit | What the record does not prove | It stops an example becoming a false claim | The system that holds this log, who may write to it and how long entries are kept is not published yet, so we describe the method rather than name a tool we have not settled on. ## We read the source, not only a summary A source can be useful and still be wrong on a detail. For a law we read the section that makes the rule. For a rule we read the setting or contract behind it. For a sum we redo the maths. We split a primary source from an account of it. The law itself is primary. A legal article can explain it, but it is not the law. Take the Digital Personal Data Protection Act, 2023. Its rules on notice and consent sit in sections 5 and 6 ([the MeitY text](https://www.meity.gov.in/static/uploads/2024/06/2bf1f0e9f04e6fb4f8fef35e82c42aa5.pdf)). The detail of what a notice must say sits in the rules made under it ([DPDP Rules 2025](https://www.dpdpa.com/DPDP_Rules_2025_English_only.pdf)). If two good sources clash, we do not pick the handier one. We go to the main text, say a point is unclear, or hold the claim back. That approach is set out in [our Content Rule](/editorial-policy). ## How we calculate overround from decimal odds Overround is the fee built into a set of odds. It is not shown as its own line. To find it, turn each decimal price into an implied chance. Add those chances up. Then take away 100%. The first formula is: **implied probability = 1 ÷ decimal odds**. The second is: **overround = sum of implied probabilities − 100%**. A fair market with two results would add up to 100%. **Illustrative two-outcome calculation. This is not an operator measurement.** - Input: one outcome is 1.90 and the other outcome is 1.90. - So the first chance is 1 ÷ 1.90 = 0.5263, or 52.63%. - So the second chance is 1 ÷ 1.90 = 0.5263, or 52.63%. - Sum: 52.63% + 52.63% = 105.26%. - So the overround is 105.26% − 100% = **5.26%**. A fair pair would be 2.00 and 2.00. Each price implies 50%, the total is 100%, and the overround is zero. Assumption: this case has two results only. A market with more results needs every one of them in the sum. Miss one and the sum is wrong. You can redo the same steps with a calculator. Our [overround guide](/guides/what-is-overround) adds context. [Our guide to how odds are set](/guides/how-bookmakers-set-odds) shows why a price can move. ## How we show the cost of a margin An overround is a percentage. To make the scale plain, we multiply turnover by the margin. That gives an expected cost at that margin, not a promise about one event or one person. **Illustrative cost at the 5.26% example margin.** - Input turnover: ₹10,000. - Input margin: 5.26%, or 0.0526. - So the sum is ₹10,000 × 0.0526 = ₹526. - So the expected margin cost is about **₹526**. Assumptions: every bet in that turnover carries the same 5.26% margin. The result is an expected cost, not a claim about one bet. It does not depend on which side a person picks in the two-outcome case. We use this case because the inputs are open to view. It does not measure a live operator's margin. A real measure would need a dated list of every price in a named market, with the time we saw them. ## How we estimate the real value left after a wagering rule A bonus has a face value and a cost to use it. The key input is the wagering rule. We multiply the required base by the stated multiple to get turnover, then multiply turnover by the example margin to get the expected cost. **Case A: a ₹1,000 illustrative bonus with a five-times rule.** - Bonus face value: ₹1,000. - Wagering rule: five times the bonus. - So the turnover needed is ₹1,000 × 5 = ₹5,000. - So the cost at 5.26% is ₹5,000 × 0.0526 = about ₹263. - So the value left is ₹1,000 − ₹263 = about **₹737**. That is about 74% of face value. **Case B: ₹1,000 deposit plus ₹1,000 illustrative bonus with a forty-times rule on both.** - Base: ₹1,000 + ₹1,000 = ₹2,000. - So the turnover needed is ₹2,000 × 40 = ₹80,000. - So the cost at 5.26% is ₹80,000 × 0.0526 = about **₹4,208**. - That is more than four times the face value of the bonus. Assumptions: the 5.26% margin comes from the teaching case above, not a measured market. We assume every unit of turnover meets the minimum-price rule. Time limits, maximum stakes and market limits are left out. Each one can cut the real value. These are teaching sums, not offers. For terms, see [Wagering Requirements](/guides/wagering-requirements-explained) and [Expected Value in Betting](/guides/expected-value-explained). ## What we do not measure, and why We build no operator ratings. A rating would need named operators, weights and a public tip on where to play. We do not measure a person's chance of winning. A cricket result is not fixed, and an odds number is not a promise. Nor do we judge an app safe from a logo, a review count or one claimed licence. We do not test deposit routes, payout routes or APK files. Those steps could help a reader use an online money game. Our page on [APK sideloading risks](/guides/apk-sideloading-risks) covers the risks without linking to any file. ## A reader can reproduce a check To redo a source check, open the source and find the line we name. Compare it with our wording. To redo a maths check, use the listed inputs and formula. If your result differs, send us the steps and the page URL. A live-looking number with no date cannot be redone. Dates on this site are IST, written as day, month and year. How the log is stored and versioned is part of what is not published yet. Raise a concern through [Contact Cricketin](/contact). Errors we confirm go into [the corrections log](/corrections). ## Questions people ask ### How do you calculate overround? Turn each decimal price into a chance with 1 divided by the odds. Add the results. Then take away 100%. Every result in the market must be in the sum. ### Is the 5.26% figure a live operator result? No. It is a teaching case, and we label it as one. It is built on two prices of 1.90. It measures no operator. ### Why do you not publish operator ratings? A rating would name operators and tip them to you. We run no rankings, no reviews, no operator links and no ads for online money games. ### Can I check your calculation myself? Yes. Use the input, the formula and the result we give. If you spot a slip, send your steps with the page URL through the verified contact route.