Scope: What We Tested And What We Did Not

This page is the highest-risk one on the site for overclaiming in either direction, so the boundaries go first rather than in a footnote.

Every probability quoted below is textbook combinatorics from a fair deck. None of it is measured site data, and it should not be read as any. If the deal were not fair, these figures would describe what should happen, not what did.

What "Rigged" Would Have To Mean

"Rigged" is used to mean four quite different things, and only one of them is an accusation about the cards.

Only the last row is testable from outside, and it is a redemption question rather than a card question. We track it separately in the payout timeline guide.

Rake Is Charged On Pots, Not On Winners

This is the load-bearing fact and it is worth stating plainly: in a cash game the house takes its cut out of the middle of the pot before it is pushed. It does not matter who drags it. Two players contest a pot, the rake comes out, and the operator's revenue for that hand is fixed the moment the pot reaches a given size.

On Global Poker that cut is 5% of the pot, capped at 5 SC, with no flop no drop — a hand that ends preflop is raked nothing at all. Tournaments carry a 10% fee on the buy-in, and the programme returns 30% as rakeback. The arithmetic of what that costs you per hundred hands is worked through in the rake math guide.

Now ask what a rigging engine would buy. Making one player win more does not increase the rake on a single pot. Making pots bigger does — but only up to the cap, above which extra action is revenue-neutral to the house and pure risk to the operator. The variable that genuinely drives revenue is hands dealt per hour, which depends on players continuing to sit down. A rigging scandal is precisely the event that empties the tables, which is why the incentive runs toward a deal nobody can credibly complain about.

The Combinatorics Of A Bad Beat

A set losing to a flush feels impossible. It is not close to impossible. The following are computed from a standard deck by counting cards, and they are the same numbers in any textbook — they are not measurements of this site's hands.

Read the third row again. A player who flops a flush draw and sees both remaining cards gets there roughly a third of the time. That is not a rare event dressed up as a bad beat; it is a coin flip with slightly worse odds than a coin. Your set is a favourite, and favourites lose about as often as the maths says they do.

What we deliberately will not do is multiply these together into a single "how often will my set get cracked" figure. That joint probability depends on how frequently opponents actually hold draws and how often they call, which is behavioural, table-dependent, and not something combinatorics can supply. Anyone quoting you a clean number for it has invented the inputs.

The one-in-forty-four river card in the last row is the honest source of the feeling. It happens. At a rate of about twice per hundred opportunities, and you will not remember the ninety-eight.

Why Online Feels Worse: Hands Per Hour

Nothing about the deck changes online. What changes is how many times per hour you ask it a question. The ranges below are the conventional figures cited across the poker world, not measurements we took:

That is roughly an order of magnitude. A player who moves from a weekly live game to four online tables does not start seeing improbable hands more often per hand — they start seeing them about ten times more often per hour, and roughly a hundred times more often per week once session length rises too. The frequency of the event is constant. The rate of exposure is not, and the human memory that flags "this cannot be random" is calibrated to the live rate.

Add the second effect: online you see every showdown, logged, replayable, with the losing hand shown. Live, most of those hands muck face down and you never learn you were beaten by a two-outer.

Monte Carlo Survival Simulator

Bad beats are the story players tell. Downswings are the thing that actually breaks them. Below, a simulated player with the win rate you specify plays the hands you specify, thousands of times over. Nothing in this model can cheat — it is a normal distribution and a seeded generator — and it still produces stretches that look exactly like being cheated.

Reading The Simulator

Run the defaults. A player winning at 2.5 bb/100 — a genuine, respectable, better-than-most win rate — still finishes 50,000 hands below where they started a substantial share of the time, and the average worst drawdown along the way runs to many buy-ins. At four tables that 50,000 hands is a few months of serious play. Months. Of losing. While winning.

Now drop the win rate to zero and watch what stays the same. The drawdown figure barely moves, because drawdown is driven almost entirely by standard deviation, not by edge. This is the point the whole page turns on: the losing stretches that feel like proof of cheating are produced by the variance term, and the variance term does not care whether you are good.

Two honest limits on this model. It uses a normal approximation per 100-hand block rather than dealing cards, which is accurate in aggregate but understates the fattest tails — real poker results are slightly more extreme than the normal curve at the edges. And it assumes your win rate is a constant, when in reality it moves with game selection, tilt, and the quality of the players who happen to sit down. Both simplifications make the simulator's downswings optimistic.

The Complaints That Are Actually Real

Rigging is the label; these are usually the underlying grievances, and three of the four are legitimate.

• Downswing length. Covered above. Real, universal to poker, and not evidence of anything about a specific site.
• Table and game selection. A smaller player pool means fewer soft tables and a higher share of regulars. Your win rate against a table of competent grinders is a different number from your win rate against recreational players, and if the pool composition shifted, your results shifted with it. This is a real, non-conspiratorial reason results can decay without anything being wrong with the deal.
• Rake drag on a marginal win rate. The most under-appreciated one. A 5% cut with a 5 SC cap is a steady subtraction from every raked pot you win, and at low stakes it is proportionally largest. A player whose pre-rake edge is small can be a genuine winner at the tables and a loser on the balance sheet. Nothing is rigged; the arithmetic simply does not clear. Run your own numbers in the rake math guide before concluding the cards are the problem.
• Redemption friction mistaken for refusal. A first cashout held for identity checks reads as the site keeping your money. It usually is not — see the payout timeline guide for what each phase is doing.

One structural note on this last point: because the SC and GC currencies behave differently, a player who has been grinding the wrong one can hit the minimum far later than expected and read the delay as obstruction. The SC vs GC guide covers which balance is the one that matters.

What Would Actually Constitute Evidence

We are not asking anyone to take the fairness of a deal on trust. We are saying the standard of proof for "rigged" is well defined and nothing published clears it. Here is what would:

Individual anecdotes cannot reach this bar at any volume, and the reason is not that players are unobservant. It is that a session is a sample chosen after the surprising hands have already occurred, and no amount of such samples adds up to an unbiased one. Ten thousand people each posting their worst beat produces ten thousand genuine bad beats and zero evidence.

If a study meeting those four conditions is published and shows a deviation, we will update this page and say so. That is the condition under which our position changes. Until then, the honest statement is the narrow one: no public evidence supports the claim, and variance explains the reported experience without needing anything else.

The Terms We Can Verify

Card fairness is not checkable from outside. The commercial terms are, and these are what our tracker records:

None of that proves the cards are fair. It does establish that the operator publishes terms, pays through named channels on a tracked schedule, and blocks registration where its model is restricted — which is the profile of a business protecting a rake stream, not one planning to abscond with it.

Bottom Line

The rigging accusation asks the operator to take on existential legal and commercial risk in exchange for revenue it already collects either way. Meanwhile the deck, dealt honestly, produces flush draws that get there a third of the time, one-outers a couple of times per hundred, and downswings measured in months for players who are genuinely winning. Every symptom is accounted for before you need a conspiracy.

Our position is deliberately limited: we did not audit the RNG and we are not certifying anything. We are saying that no evidence in public supports the claim, that the incentive structure argues against it, and that variance explains what players report. Bring a reproducible million-hand analysis showing otherwise and this page changes.

Next Steps

Work out whether rake is quietly eating your edge in the rake math guide, sort out which balance actually cashes out in SC vs GC, check the withdrawal schedule in payout times, weigh the alternative in Global Poker vs ClubWPT, or read the full platform breakdown on the Global Poker review. Everything else lives in the guides index and the tools directory.