Methodology
How we measure each criterion, in enough detail that you can re-run the test.
We score each platform on four criteria and combine them with a fixed weighting: KYC 0.20, RNG 0.35, table limits 0.20, support 0.25. The weighting is not negotiable and not platform-dependent — every platform is graded against the same rubric. RNG receives the largest weight because no amount of skill compensates for an unfair deck; if the RNG is biased, the game is lost before the first card is drawn. The remaining three criteria are weighted roughly equally because they trade off against each other: fast KYC is worth little if a withdrawal dispute takes three days to answer, and excellent support is worth little if the only available tables are beyond your bankroll.
For each criterion we publish the sample size, the time window, the inclusion rule for outliers, and the way we aggregate. Where the methodology differs across criteria, the per-criterion notes below call that out. We do not publish the raw tick-by-tick data because the sample is small enough that a single operator's marketing team could identify and dispute the underlying tickets; the methodology is published in enough detail that a reader with a stopwatch and a basic stats package can replicate the scoring from scratch.
All measurements were taken between Q1 and Q2 of the current cycle. Each platform's data set covers a rolling six-week window within that period, with the start date offset per platform to avoid sampling bias (we did not measure every platform in the same week). A platform that substantially changed its verification flow, RNG certificate, or support vendor mid-window is flagged in the row's footnote rather than re-scored silently — re-scoring mid-cycle is the kind of methodological drift that makes a comparison ledger useless six months after publication.
How we measure KYC flow
KYC is scored on the median minutes from account creation to first real-money hand, measured across a 25-signup sample per platform. Each signup uses a fresh device, a fresh PAN, a fresh mobile number and a fresh email address; we do not reuse identities across platforms because the verification rules are platform-specific. The 25-signup sample is split: 15 signups use only the documents the platform marks as "default" (typically PAN), 10 signups add a second document (typically Aadhaar) at signup time. We record the time at each of four checkpoints: account creation, document upload complete, auto-decision rendered, first real-money table joined.
The score aggregates those 25 signups by computing the median end-to-end time (from account creation to first real-money hand) and mapping it to the published scoring bands: 9–10 for under five minutes, 7–8 for under thirty minutes, 5–6 for under twenty-four hours, 3–4 for twenty-four to seventy-two hours, 0–2 for longer than seventy-two hours or for signups that are never auto-approved. We then apply a small penalty (–1 score point) if the platform requires mandatory video verification at signup, and a small bonus (+1 score point) if the platform publishes a public SLA for verification turnaround.
The weighting rationale for KYC at 0.20 is that verification speed matters primarily as a user-experience signal and secondarily as a fraud-control signal. A platform that takes twenty-four hours to verify is not necessarily unsafe — a manual review with a trained risk team is often safer than a five-minute auto-approve. But fast verification is a measurable proxy for operational maturity, and a slow verification flow typically correlates with a slow support flow on the same platform, so KYC has predictive value beyond its face meaning.
How we measure RNG disclosure
RNG is scored on the strength of the independent audit certificate, not on the platform's marketing claim. The audit must be from a recognised testing laboratory (iOLE, GLI or BMM Testlabs), must name the specific game version covered, must name the date range of the audit, and must be verifiable on the lab's own public registry. We award 10 points for an iOLE certificate with quarterly cadence and a publicly accessible certificate ID, 8 points for a GLI certificate with semi-annual cadence, 7 points for a BMM Testlabs certificate with annual cadence, and 6 points or below for any audit that cannot be located on the lab's registry within five minutes of search.
The sample window for RNG is the most recent twelve months — a certificate older than twelve months scores a full point lower than a current certificate, because the lab's stamp only attests to the version that was tested. If a platform changes its RNG implementation, redeploys its server, or ships a new game variant without re-certifying, the prior certificate does not extend to the new build, regardless of marketing language. We re-verify the certificate date on every scoring run, which means a platform's RNG score can move down even when nothing about the platform itself has changed.
The weighting rationale for RNG at 0.35 is straightforward: the RNG is the only criterion that touches the integrity of the game itself. A platform with poor KYC is annoying; a platform with a biased RNG is rigged. We chose 0.35 — the largest single weight — because no other factor in the ledger has the same first-order effect on whether a winning hand is a winning hand. If you only check one criterion before depositing, check this one.
How we measure table limits
Table limits are scored on the breadth of the stake range that is publicly available to a non-VIP, non-invite-only account. The score is the count of distinct stake tiers reachable from the lobby without an account upgrade: we count micro-stake (under ₹10/point), low-stake (₹10–₹50/point), mid-stake (₹50–₹500/point) and high-stake (above ₹500/point). A platform with all four tiers scores a 9; a platform with three tiers scores a 7; two tiers scores a 5; one tier scores a 3. We then deduct two points for any "VIP-only" or "invite-only" high-stake table, because those tables make the stake distribution unverifiable and tend to obscure the rake.
The sample window for table limits is a single seven-day observation per platform, taken at the same time of week to avoid weekend-versus-weekday bias. We capture a screenshot of the lobby at each of four checkpoints across the seven days, then resolve any disagreement between checkpoints by visiting the platform a fifth time and recording the modal tier list. The sample is small but the underlying variable — the lobby tier list — is itself small, so the small sample is not a methodological weakness: the lobby tier list is bounded by what the platform chooses to publish.
The weighting rationale for limits at 0.20 is that the stake range matters most for bankroll calibration. A reader with a ₹2,000 monthly budget needs access to micro-stake and low-stake tables; a reader with a ₹50,000 monthly budget needs access to mid-stake tables. Neither reader benefits from a stake range that does not include their tier. We weight limits at 0.20 — equal to KYC — because both are operational signals that affect everyday usability, but neither has the game-integrity weight of RNG.
How we measure support response
Support is scored on the median minutes from ticket submission to first human-replied message, measured across a 30-ticket sample per platform. Tickets are submitted at randomised times across a rolling fourteen-day window, with each ticket representing one of five categories: KYC document rejection, withdrawal delay, table dispute, account lock, and bonus-credit discrepancy. We do not submit duplicate tickets on the same category within the same window, because platforms can and do correlate repeat tickets within an account. We also do not submit tickets that include real money or real KYC documents — every ticket is a synthetic test of the response pipeline.
The aggregation is the median, not the average, because the average is distorted by a small number of tickets that wait days for a reply. A platform with a four-hour median and a seventy-two-hour tail (one ticket unanswered in three days) is reported honestly as a four-hour median with one outlier; a platform with a four-hour average and a thirty-hour median is reported as a thirty-hour median. The published scoring bands reward medians under two hours with a 9–10, medians of two to six hours with a 7–8, and so on, with a –2 point penalty applied to platforms where more than 5% of submitted tickets receive no reply within the seven-day observation window.
The weighting rationale for support at 0.25 is that support is the criterion that converts a small problem into a small annoyance rather than a small problem into a lost deposit. A platform with a four-hour support median is materially less risky than a platform with a twenty-four-hour support median, even when the two platforms have identical KYC, RNG and limits. We weight support slightly higher than KYC and limits because it is the criterion that activates when the other three have failed: a KYC issue requires support to resolve, a withdrawal delay requires support to escalate, a table dispute requires support to adjudicate.
