What is Casino Game Math & RNG Certification

Key Takeaways

• RNG certification for casino games is the process of defining a game’s probabilities, RTP, and volatility, then having an accredited lab verify that its random number generator produces fair, unpredictable outcomes that match those numbers.
• The math model sets reel strips, symbol weights, paytable values, and bonus odds to hit a target RTP, such as 96%, within limits the jurisdiction allows.
• Labs like GLI, eCOGRA, iTech Labs, and BMM Testlabs test the RNG’s source code and statistical output, then run millions of simulated rounds to confirm real payouts match the declared RTP.
• Certification isn’t a one-time stamp, since labs keep monitoring after launch and code changes can trigger a re-test. 
• Learn how Idea Usher can help businesses develop a certification-ready casino game that meets lab and regulatory requirements from the start.

Before a casino game goes live for real-money players, an independent test lab has to examine it. GLI and BMM Testlabs are two of the better-known labs that do this, and regulators in licensed markets generally want a report from a lab they recognize. The examination has two parts. In the RNG review, the lab runs statistical tests on the generator’s output and reads its code, looking for bias or patterns that would let someone predict a slot spin or a card shuffle. In the math review, the lab rebuilds the game’s probability model from the paytable and reel data.

It then checks that the RTP and hit frequency the operator claimed match what the math produces, and that the house edge stays within the limits set by the jurisdiction. A game that fails either part goes back to the operator for changes and cannot launch until it passes. In this blog, we’ll discuss what casino game math and RNG certification involve, how the process works step by step, and where games most often go wrong.

Why Does a Casino Game Need RNG Certification?

A casino game needs RNG certification because software decides every spin and card, and players can’t see inside it. An independent lab tests the game to prove the results are random and can’t be predicted or changed. In the UK, new products must be tested by an approved test house before release. According to Technavio, this matters more as the market grows. The global casino market is forecast to grow by USD 34.20 billion between 2025 and 2030, at a CAGR of 3.4%. More regulated markets and easier licensing are driving that growth, and every regulated market wants proof that its games are fair.

Source: Technavio

Fair and Unbiased Game Outcomes

Fair means every result has the exact chance it is supposed to have. A lab doesn’t judge whether a game feels fair. It runs the generator through millions or even billions of outcomes and applies standard statistical tests. If some numbers show up too often, or one result hints at the next, the game fails. Labs also simulate many millions of rounds to check that the real payout matches the RTP the maker promised. 

What Labs Test

• Chi-square: do results show up as often as they should
• Serial correlation: does one result give a hint about the next
• Runs test: are there odd streaks that point to bias
• Gap test: how far apart a certain result appears 

Lightning Roulette Example

Evolution’s Lightning Roulette shows why this matters. It mixes a European roulette wheel with randomly generated multipliers. Evolution calls the Lightning games the first online game shows to blend live and RNG play. A player can watch the wheel spin but can’t see how the multipliers are picked, so a lab has to test that part. 

Evolution’s newer Lightning Storm has a 39-segment DigiWheel and a top prize of 20,000x the bet. Evolution also brought its live games to Delaware through a partnership with BetRivers. 

Regulatory and Compliance Requirements

In the UK, a game can’t just launch and hope for the best. The Gambling Commission wants the test house’s game and RNG report sent in after testing is done but before release. Its rule on random outcomes says results must be “acceptably random.” Other rules cover how results get decided and how live dealer studios are checked. Operators have to keep this paperwork ready. 

Pragmatic Play’s Gates of Olympus shows how one change touches the math. In the original slot, multiplier symbols land with random values between 2x and 500x. The newer Gates of Olympus Super Scatter adds a Super Scatter symbol and a 50,000x max win. Its bonus starts with 15 free spins, and the update came alongside Bandit Megaways and Big Bass Bonanza 1000. Some markets also get two bonus buy options. Every one of those numbers changes the odds, so a lab has to look at the new version again.

Player Trust Through Verified Randomness

Players can’t check a spin themselves, so they have to trust someone. The Gambling Commission found that 33% of people surveyed in 2017 believed gambling is fair and can be trusted, down from 48.8% in 2008. The Commission’s research says the main thing that decides where people gamble is a company’s name for being fair and trustworthy. A lab seal gives players something real to point to.

Trust breaks fast when a generator gets tampered with. Eddie Tipton, the security chief at the Multi-State Lottery Association, hid code in random number software and rigged at least five draws worth more than $24 million in Colorado, Iowa, Kansas, Oklahoma and Wisconsin. Investigators found the code was added after a security firm had already audited the machine. That is why labs read the source code. Most labs also keep monitoring after launch. It was a lottery and not a casino, but the lesson is the same. 

What Makes Casino Game Mathematics Certification-Ready?

A game’s math is certification-ready when it is fully written down and the game really does what the papers say. The probability model, RTP, reel strips, paytable and bonus rules all have to match each other and match the code. A lab should be able to rebuild the numbers from the files you send. If its own simulated rounds land on the same figures you claimed, the math passes. If one number drifts, the game goes back for fixes.

1. Probability Models Behind Game Outcomes

Every game starts with a model that lists each possible outcome and its chance. Baccarat is a clean example. The Banker hand wins about 45.86% of the time, the Player hand about 44.62%, and a tie about 9.52%. Those chances decide everything that comes later. A lab checks that the game code produces the same chances as the model. Regulators also set limits on the model itself. New Jersey says a slot can’t offer a play with odds worse than 100 million to 1. 

What a lab looks at in the model:

• Every outcome and the chance of each one
• The payout for each outcome
• How often bonus features trigger
• For games like video poker, an RTP that assumes the player uses the best strategy 

2. RTP, House Edge and Expected Returns

RTP is the share of money a game pays back over a very long time. The house edge is whatever is left, so the two always add up to 100%. Short sessions can look very different. A game’s actual return should move toward its RTP over millions of spins. Regulators also set a floor. Nevada requires a theoretical payout of at least 75% and New Jersey requires at least 83%. 

Mega Moolah Example

Mega Moolah from Games Global shows how RTP can get tricky. Its base game RTP is 88.12%, and part of every stake funds four progressive jackpots. That is low, but the jackpot money counts too. In New Jersey the expected contribution from a progressive award can count toward the minimum RTP. The jackpot has also made headlines. Mega Moolah paid out €19.4 million in April 2021, the largest single online jackpot win at that time. 

A Betway player later hit €13.5 million ($14.7 million) on Book of Mega Moolah. Games Global’s jackpot network has also opened to SkillOnNet players in the UK, Mexico and Sweden. 

3. Reel Strips, Symbol Weights and Paytables

A modern slot doesn’t really spin anything. The RNG gives out numbers and the game maps them to virtual reel positions. The reels just show the result. Each symbol gets a number of stops on the strip, and rare symbols get fewer. These weights and the paytable are set around a target RTP and volatility. So a lab needs the exact strips and weights. Moving one stop on one reel can change the RTP. The UK rules also say any scaling of the RNG output must keep it random. 

4. Hit Frequency and Volatility Distribution

Hit frequency is how often a spin pays anything at all. Volatility is how big and how uneven those wins are. Two slots can share the same RTP and still feel very different. A high-volatility slot at 96% may pay rarely but big, while a low-volatility one pays smaller amounts more often. Volatility changes the size and timing of payouts, not the long-term return. A lab checks that the spread of wins matches what the developer filed. 

Starburst Example

NetEnt’s Starburst is a good low-volatility case. It runs at 96.09% RTP with low volatility, and some operators run lower RTP versions. Its hit frequency is listed at 22.7%. NetEnt now sits inside Evolution’s portfolio. Starburst XXXtreme came out in 2021 and adds Random Wilds and XXXtreme Spins. A new feature like that changes the hit rate, so the new game needs its own math file.

How to Calculate and Validate Casino Game RTP?

To calculate the RTP of a casino game, you take every possible win, multiply it by how often it can happen, and add all of those numbers up. That total is the theoretical RTP. To validate it, you run a huge computer simulation and later compare real play against the same number. Labs and regulators look at both steps, so a game that skips one can get stuck before launch.

1. Theoretical vs. Actual RTP

Theoretical RTP is the designed return of the game, and actual RTP comes from the win and turnover figures of the live game. You get actual RTP by dividing wins by turnover, and the allowed tolerance is wide when only a little play has been measured. A game with 96% RTP gives back $96 for every $100 on average. A short run can still look far off and be completely normal. 

The UK Gambling Commission gives an example of a game designed at 91.68% that reached 90.42%. Its volatility allowed a tolerance of about 1.75 points either side after 400,000 plays, so it was fine. The table shows how that gap gets smaller as more games are played. 

These figures use a 95% confidence level and a game with a volatility of 5.6. 

2. RTP Across Base and Bonus Games

GLI splits a game’s RTP into the contribution from the base game, the feature games and the jackpots, and it also measures volatility. You work out each part by multiplying its chance of happening by its average win. For a bonus round, you also multiply by how likely it is to trigger. Then you add the parts together, and the total has to match the RTP printed in the game rules. 

Total RTP = base game + feature games + jackpots

Gates of Olympus Super Scatter

Pragmatic Play’s Gates of Olympus Super Scatter shows why each part needs its own check. Its Free Spins mode has a 96.5% RTP and costs 100x the bet to buy. Its Super Free Spins mode has 96.49% and costs 500x. Pragmatic Play also made 95.5% and 94.5% versions. So the RTP a player gets depends on the setting the casino picked. 

Pragmatic Play launched the game on 28 April with a Super Scatter symbol that can pay up to 50,000x. On 28 September, it released Gates of Olympus 2500, which has multipliers up to 2,500x and four free spin modes. It also made Gates of Olympus POP for its new casual sub-brand, POP, which started on 31 July. 

3. Paytable Validation Through Probability Models

To validate a paytable, you count every outcome and check that the wins add up to the RTP on paper. Course notes on slot math show this with the Mills Five Cent Bell, which has 20 stops on each of three reels. That gives 8,000 outcomes, and 1,073 of them pay, so the hit rate is 13.41%. The wins add up to 7,556 out of 8,000, which is a 94.45% return. 

Lightning Roulette example

Evolution’s Lightning Roulette gets the same check on a table game. Before each spin, one to five numbers get a multiplier from 50x to 500x. The straight-up payout drops from 35:1 to 29:1. At that payout, a straight-up bet only returns 81.08%. The multipliers therefore have to add back about 16 points. Evolution lists a 97.10% RTP for straight-up bets and 97.30% for all other bets. 

Evolution has kept adding to the family. XXXtreme Lightning Roulette brought Chain Lightning and Double Strike multipliers that can reach 2,000x. Lightning Storm added five bonus games with a top prize of 20,000x. At ICE in January, Evolution showed MONOPOLY Roulette, one of several upcoming Hasbro titles. 

4. Monte Carlo RTP Verification

Monte Carlo simulation means playing a game millions of times on a computer with random numbers and then averaging the payouts. GLI mathematicians write programs that check the pay combinations from the developer’s payout table, and they use simulators when there are too many combinations to count. The simulated number is then compared with the math answer. If the two do not match, something in the model or the code is wrong. 

More spins make the answer tighter, but only slowly. In the table above, 100,000 plays leave a gap of 3.47 points and 1,000,000 plays leave 1.10. At a 95% confidence level, a healthy game can still miss the range about 1 in 20 tests. The Commission says one failure does not prove a fault, but repeated failures might. 

• Work out the exact RTP from the paytable.
• Simulate millions of rounds and compare the average with that number.
• Check live play after launch. One method the UK Gambling Commission describes is a daily check on the last 30 days of play. 

How Does RNG Testing Work in Casino Games?

RNG testing works by proving that a game’s random numbers cannot be predicted and do not favor any outcome. A lab reads the code, runs the output through statistical tests and checks how the numbers become game results. If one of those steps fails, the game does not get certified. Any later change to the generator sends it back for testing.

1. PRNG Algorithms, Seeds and Entropy

Most casino games use a pseudorandom number generator, which is a formula that stretches a small starting value into a long stream of numbers. That starting value is the seed. The output of a generator like this is only as unpredictable as its seed. The unpredictability of a seed is called entropy, and it usually comes from operating system sources or hardware generators. GLI checks the period of the generator, the range of its numbers, and how it is seeded and re-seeded. 

Aviator by Spribe

Spribe’s Aviator shows how seeds can be shared with players. Its crash point comes from a private server seed combined with client seeds from the first three players who bet in a round. The server seed is committed as a SHA-256 hash before the round opens, then the seeds go through SHA-512 to make the result. After the round, players can check that the numbers match. The game runs at a published 97% RTP that iTech Labs and BMM Testlabs have certified. 

Spribe keeps adding to the game. Aviator Challenges is a gamification feature that lets operators set missions for their players. It was tested in Africa first and then launched globally. In June the game was added to the Games Valley aggregation platform. Spribe also has a deal with TKO that puts the Aviator logo at UFC and WWE events. 

2. Mapping Random Numbers to Outcomes

After the generator makes a number, the game has to turn it into an outcome. That might be one of 37 roulette pockets, a symbol on a reel or a card in a deck. Mistakes in this step can bias a game even when the generator is perfect. Taking a number modulo 52 to pick a card adds a small bias when the generator’s range is not an exact multiple of 52. The fix is to throw out out-of-range values and draw again. 

A famous failure shows how bad it can get. A real deck has about 2^226 possible shuffles. PlanetPoker’s shuffle used a 32-bit seed, so it could make only just over 4 billion decks. The seed came from the server clock, and syncing with that clock cut the options to around 200,000. The code also had an off-by-one error that kept the last card from staying in the last spot. Many online poker sites now use the Fisher-Yates shuffle. 

Bonanza Megaways by Big Time Gaming

Bonanza Megaways has two mapping jobs to test. The number of symbols on each reel changes every spin, and with seven symbols on all six reels there are 117,649 ways to win. So the game has to turn random numbers into reel heights as well as symbols. It lists a 96% RTP.

Bonanza Falls is the follow-up. It adds the Megadozer, a coin pusher mechanic inspired by arcade games. Big Time Gaming and Light & Wonder released it together in the USA, Canada, and EMEA. Its max win is 93,540x, compared with 26,000x for the original. 

3. Statistical Tests for Randomness and Bias

Labs run big batches of a generator’s output through test suites. GLI uses the DIEHARD battery and outcome distribution tests. NIST’s SP 800-22 suite has 15 tests, and at a 0.01 significance level, about 1% of good sequences are expected to fail each one. So one failed test is a flag to look closer. It does not prove the generator is broken. 

• Chi-square checks that outcomes are spread evenly.
• The runs test looks for streaks or patterns that should not be there.
• Serial correlation checks that one result does not depend on the one before it. 

Passing the tests is still not enough. A generator can pass test batteries and still be predictable if its seed is weak or its state is exposed. The Mersenne Twister is a good example. It passes the Diehard tests, but an attacker who sees 624 outputs can predict every later one. That is why labs also look at seeding and code. 

4. RNG Source Code and Algorithm Verification

A lab also reads the code that makes the numbers. GLI reviews the RNG source code and must confirm that this code was used to build the game, either by checking signatures or by watching the build get compiled. It also looks at the period, the range, the seeding, and the background cycling of the generator. 

Code checks matter because code can change after approval. Investigators in the Tipton lottery case found extra code in a recovered generator that was added after a lottery security firm had reviewed and approved it. The code stopped the generator from making random numbers on three days of the year when they fell on certain weekdays. Eddie Tipton was convicted of fraud over an attempt to claim a $16.5 million Hot Lotto jackpot. GLI’s conclusions stop applying after changes unless those changes are also examined. 

What Do Casino Game Testing Labs Actually Check?

Testing labs check four things. They look at the code that makes the random numbers and test whether the results really look random. Then they check that the math behind the payouts is right. Last, they make sure the game follows its own rules every time. A game has to pass all four before it can go live, because passing one does not cover the others.

1. RNG Source Code Review

A lab starts with the code, not the game screen. The reviewers read how the random number generator is built and ask if anyone could guess what comes next. GLI compiles the submitted source code on its own and takes a checksum, so the lab knows the code it read is the same code running in the game. If the studio changes even one file later, the lab has to look again. 

GLI’s RNG reviews can include the RNG period, seeding and re-seeding, background cycling, and cryptographic security. In plain words, the lab wants to know: 

• how long the generator runs before its pattern repeats
• how the starting value (the seed) is picked and refreshed
• whether the generator keeps running between spins
• whether an attacker could work out the next number

Why seeding matters

Most game RNGs are pseudo-random. That means they follow a formula, and the seed decides everything that comes out. If the seed is weak or the inner state can be rebuilt, the whole generator is basically broken. GLI-19 puts a lot of weight on resisting this kind of attack, which is why seeding gets its own line in the review. 

2. Statistical Randomness Testing

After the code, the lab collects a huge pile of output and runs it through test batteries. Labs run this across millions to billions of simulated outcomes. The idea is simple. If the generator is fair, no number should show up more than it should, and no hidden pattern should exist. GLI-19 asks the RNG to pass recognised statistical tests at a 99% confidence level. GLI and eCOGRA both hold ISO/IEC 17025 accreditation, so their testing methods get checked too. 

The UK regulator keeps the bar short and clear. Random number generation and game results must be “acceptably random.” Passing the numbers is how a lab proves that. 

3. RTP and Paytable Verification

RTP is the share of all bets a game pays back over a very long time. The lab takes the paytable and the game math, builds its own model, and simulates play. Then it checks if the RTP printed in the help screen is the real RTP. This matters to regulators because players can lose money unfairly when the real number is off. The UK Gambling Commission lists unfair financial loss from an RTP that is not in line with the expected value as a risk, and it asks for testing before release and yearly game testing audits. 

Example: Mega Moolah

Mega Moolah from Games Global is a good case because the jackpot makes the math harder. One review lists its RTP at 93.42%. The jackpot grows with every bet, so the lab has to check the base game and the jackpot part. UK rules say the jackpot RTP can be shown as one combined figure or split into base game and jackpot. The game has a lot at stake. Mega Moolah and WowPot titles paid out more than €155m ($168.9m) in a single year. 

The brand keeps growing, and every new version needs its own check. Absolootly Mad Mega Moolah from Triple Edge Studios mixes Rolling Reels and a multiplier trail with the four-tier jackpot. SkillOnNet also launched the Games Global jackpot portfolio in the UK, Mexico and Sweden, and Games Global unveiled a new progressive network called Jackpot Madness. A new market or a new game wrapper does not skip the math check. 

4. Game Rules and Outcome Consistency

The last check is about whether the game does what it says. The lab plays through every bet type, bonus trigger and payout and compares each one with the written rules. The game art and text must give the player enough information to know if they won or lost and how much. If a virtual game copies a real object, the odds have to match too. A coin must land heads with a probability of 0.5 every time. 

Getting this wrong can cost money. The UK Gambling Commission fined the operator of Betfred.com £240,000 for slot features that broke its technical standards. 

Example: Lightning Roulette

Lightning Roulette from Evolution is tricky because it is half real and half computer. It uses a normal roulette wheel but adds RNG “Lucky Number” wins in every round, shown as lightning strikes. So the lab has to check two things that must never touch each other: the wheel result and the random multiplier. One cannot change the other.

The game also keeps spreading. Evolution made a deal with Scientific Games to bring Lightning Roulette to land-based casinos. Betway XXXtreme Lightning Roulette in South Africa adds Chain Lightning Numbers and Double Strike Multipliers with multipliers up to 2,000x. Every extra feature like that is one more rule for the lab to test. 

5. Platform and Software Integrity

Passing in the lab is not enough. The game that goes live must be the exact game that was tested. GLI reports list digital signatures like MD5 and SHA-1 hashes for the tested files. A regulator can compare those against what runs in production. Labs also test the platform around the game. GLI-19 includes rules for the system clock, the control program, player location, reporting, and disabling games. The work does not stop at launch either. Most labs keep monitoring for compliance after a game goes live.

Is GLI-19 the Standard for Casino Game Certification?

Not on its own. GLI-19 is one of the most widely used technical standards for online casino systems, and many regulators build on it. But each regulator decides which rules apply to its market. A game may need GLI-19 or a local rulebook like the UK standards or sometimes both. So GLI-19 is a common base and not one global certificate that works everywhere.

What GLI-19 Actually Covers

GLI-19 is a standard for interactive gaming systems that regulators can adopt in full or use as a base. That means the whole platform and not just one slot. Version 3.0 was revised on July 17, 2020, and GLI offers it as a free download. A newer 4.0 draft is out for public comment. It reorganizes the chapters and expands the responsible gaming tool rules. 

The standard touches these areas:

• Player accounts, platform security, operational controls and game integrations 
• Reporting, KYC and geolocation 
• System clock and control program rules 
• A full RNG chapter and rules for disabling a game 

RNG and Outcome Requirements

GLI-19 gives the RNG its own chapter. Labs then check it in practice. In one public GLI report, the RNG went through source code review and statistical testing and passed at the 95%, 98%, and 99% confidence levels. The draft 4.0 goes a step further. It lists sections on game outcomes using an RNG, game fairness and outcome integrity, and bonus and feature requirements. So the standard cares about what the player sees and not only the raw numbers. 

Example: Aviator

Aviator is a crash game that Spribe launched in 2019. The crash point is random so RNG certification sits at the heart of it. Spribe says its titles are certified by GLI and BMM in several jurisdictions. For Italy, the games carry their own ADM RNG certificate and game certificate. The game also has a provably fair check that players can use. That check relies on cryptography and not on an outside auditor so it sits next to lab certification and does not replace it. Newer features also matter. Rain Promo drops free spins into the chat at random, and a lab has to see features like that as part of the build.

GLI-19 vs UKGC standards

The UK adds a few things of its own. The Commission expects game testing in the software and environment meant for live use. Licence holders also have to submit games test results and RNG test reports. A report written for one rulebook may not map neatly onto the other so it helps to plan for both early. 

How Requirements Differ by Market

Every market has its own mix. Some copy GLI-19 almost word for word. Michigan adopted it in Part 3 of its rules and added its own extras. Others use their own rulebook. Ontario is a good case. The AGCO does not test iGaming technology itself and instead oversees certification by registered independent labs against its Registrar’s Standards for Internet Gaming. GLI was named one of those labs in August 2021. 

• New Jersey: Play’n GO needed a Casino Service Industry Enterprise waiver from the DGE 
• Ontario: Certain live dealer equipment needed approval or lab certification from April 2023 
• Alberta: Each game must be tested for RTP, RNG and mechanics before it goes live.
• Italy: separate RNG and game certificates, as the Aviator case shows

Example: Book of Dead

Rich Wilde and the Book of Dead from Play’n GO shows how many approvals one game can need. It went live in New Jersey with PokerStars after the DGE approved Play’n GO’s licence. Ontario then granted a supplier licence, and Book of Dead was among the first titles offered there. Later, a bet365 deal brought more than 50 games to New Jersey, including Rise of Olympus 100 and Colt Lightning. Colt Lightning Firestorm also got a network-wide US release. Each new market and each new game meant another round of local approval. 

When Does a Casino Game Need Recertification?

A casino game needs recertification when a change could affect fairness. That usually means a change to the RNG, the game math or the rules. It also covers big updates to the platform that serves the game. Small changes like sounds or artwork often only need internal testing. In the end, the regulator and the test lab make the final call.

1. RNG and Source Code Changes

The RNG is the one part regulators watch the most. In the UK a new game or RNG can only be released after testing is done and the report is sent to the Commission. If an RNG update could touch hundreds of games, the licensee does not have to retest every single one. A representative sample is retested to show the update did not change how they work. That sample has to cover each game type and generation and not just similar titles.

A lighter path exists for small code changes. When iTech Labs reviewed an updated NetEnt RNG, it compared the new source code to the approved version. It found no change in function and the RNG still used the Fortuna algorithm. The original tests were not repeated. This shows that a lab can sign off a code change without a full retest when the logic stays the same. Whenever the scope is limited, the updated report must point back to the original one and list the changes and the testing done.

2. RTP and Paytable Updates

Changes to the math are where most recertification happens. The UK testing strategy treats changes to the RNG or scaling and mapping or the game rules as fairness changes. A paytable edit falls under the game rules. A new RTP setting follows from the math underneath it. Not every change is big, though.

Licensees decide major or minor with their test house, and they must give a reason for each call. To keep them honest, the UK requires an annual games testing audit. The auditor samples major and minor updates to check that each was sorted correctly. The game report also names the theoretical RTP and the software digital signature so a later change can be matched against it.

3. New Bonus Features and Mechanics

A new bonus feature is almost always a rule change. It adds new outcomes and new ways to pay. The draft of GLI-19 version 4.0 even gives bonus and feature requirements their own section. So a lab will test the new feature and also check that it plays well with the base game.

Example: Starburst XXXtreme

NetEnt launched Starburst XXXtreme in July 2021 as a follow-up to the classic Starburst. NetEnt is part of Evolution Group. The big new mechanic is XXXtreme Spins. Players can pay 10 times the bet for one guaranteed Starburst Wild on each spin or 95 times the bet for two. One review lists the base RTP at 96.26% and says it rises to 96.45% when XXXtreme Spins are active. The same review notes the feature is not available in the UK. That is a good reminder that one feature can pass in one market and not in another. Each version needs its own approval.

4. RGS and Platform Changes

Sometimes the game does not change at all and the system around it does. The UK says an RGS or RNG update can change the fairness of many games at once. Labs also run extra integration testing when a game is used with a different RNG or platform than the one first tested. New channels added to a game need their own test reports that point back to the original. New browser or device releases are different. The licensee tests those itself unless they touch the RGS or RNG underneath.

A simple order of work looks like this:

• Decide if the change touches fairness and mark it major or minor
• Record a change ID and the reason for the call
• Send major changes to the test house and retest a sample if many games are hit
• Submit the updated report before launch

Example: Mega Moolah

Mega Moolah from Games Global shows how platforms multiply the work. It is a progressive jackpot that runs on a shared network. When SkillOnNet added the Games Global jackpot portfolio, it went live for players in the UK, Mexico, and Sweden. That included WowPot and King Millions, besides Mega Moolah. 

The bet365 launch brought Mega Moolah and WowPot too. Games Global also released Mega Moolah 4Tune Reels. It remakes the 2006 original and lists a 90.3% RTP that includes a 5.3% jackpot contribution. Each new operator or remake means new integration checks and possibly a new report.

What Documents Are Required for RNG Certification?

RNG certification usually needs a document packet that explains how the game works and lets the lab prove it. The packet covers the RNG algorithm and source code and the math and RTP reports. It also holds the paytable and rules and the software version records. The last piece is the test certificate that comes back. Each regulator adds its own forms so the exact list changes by market.

1. RNG Algorithm and Source Code

The code is the heart of the packet. GLI asks for the final source code in full along with the compiled binaries. The code cannot still be in testing or development. GLI then builds the files itself and takes a checksum. This proves the code you sent is the code that makes the game run. Labs also want a tool that can collect the final outcomes so they can test the numbers the RNG really produces.

The UK asks for a clear RNG summary too. It covers a short description of the RNG and its version and whether it is hardware or software. It also lists the digital signature and the scope of testing and every test that was applied. Think of it as the story of the RNG on paper. A lab should be able to read it and then open the code and find the same story there.

2. Game Math and RTP reports

The math pack tells the lab what the game should do over millions of spins. A common set is the PAR sheet, the probability tables, and the feature distributions. Volatility settings and bonus trigger odds go in too. The UK game report also names the theoretical RTP so that number must match the math pack exactly.

Example: Big Bass Bonanza 1000

Big Bass Bonanza 1000 from Pragmatic Play and Reel Kingdom came out in April 2025. Its main RTP is listed at 96.51% with two alternative versions at 95.51% and 94.51%. The max win jumped to 20,000x from 2,100x in the original game. The original listed 96.71%. Big Bass Splash 1000 followed in December 2025 with 96.52% and alternative versions at 95.51% and 94.53%. Every RTP version is a different math setup. So each one needs its own numbers in the report.

3. Paytables and game Rule Specs

Studios send these along with the source code and math models. The rules file needs to be the final word. If the screen says one thing and the rules file says another the lab will flag it.

Example: Wanted Dead or a Wild

Hacksaw Gaming’s Wanted Dead or a Wild has a 5×5 grid and 15 paylines. Its listed RTP is 96.38%, and its max win is 12,500x. It has three bonus rounds, and each one has a buy price. The Great Train Robbery costs 80x the bet, and Duel at Dawn costs 200x. Dead Man’s Hand costs 400x. Its VS duels use multipliers up to 100x. All of that has to be written down exactly. Hacksaw is certified in Malta, the UK, Ontario, and Italy, and several other places. It later added a Michigan provisional licence. More markets mean more paperwork for the same rules.

4. Software Versions and test records

Labs need to know exactly which build they tested. GLI says its software check should uniquely identify the tested RNG and its source code. The packet should hold:

• Operating system and software version
• Software number and digital signature
• Channels covered by testing
• Older versions the new build replaces
• Change records with a unique change ID and a major or minor label

New Jersey also lists operating system and software version as part of the software documentation. When a game changes, the updated report should point back to the original. It lists the changes, the new testing, and the new signatures. That trail lets a regulator follow the game from the first build to the live one.

5. Independent Testing Certificates

The last document is what the lab sends back. A UK report names the test house and the supervisor who signed it off. It also gives the licensee and the date and a certificate reference.

Labs must be approved for the market you plan to enter. Lists of approved test houses differ by regulator. GLI and eCOGRA both hold ISO/IEC 17025 accreditation and the UK list uses that standard too. Check the lab before you pay for testing and not after.

How Much Does Casino Game RNG Certification Cost?

Most casino game RNG certification costs about $8,000 to $25,000 per game in estimated lab fees. A simple slot sits near the low end, and a feature-heavy game sits near the top. Retests and yearly checks cost extra. Labs rarely publish price lists, so treat every number here as an estimate and ask for a quote.

1. RNG Testing and Code Review

This is the biggest single line item. The lab reads the source code and runs statistical tests on the RNG output. It also checks that the game plays the way the documents say. One development guide puts a standard RNG slot at $8,000 to $20,000 for certification and a premium or 3D slot at $10,000 to $25,000. Another industry guide says its numbers are estimates and not published price lists. eCOGRA also does not publish a public fee schedule so any figure for it is a rough guess.

These numbers come from industry guides and not from the labs. The eCOGRA guide also lists $2,000 to $5,000 per year for ongoing monitoring. Ask each lab for a written quote that names exactly what it covers.

2. Game Math Validation Costs

Math checks get pricier as the game gets more complex. A simple slot mechanic certifies faster and cheaper than a bonus game with dozens of edge cases. One EU guide shows the gap in euros. A simple slot runs about €2,000 to €5,000 and takes 2 to 4 weeks. A complex slot with several bonus features or progressive jackpots runs about €5,000 to €15,000 and takes 4 to 8 weeks. The math model also has to be final before certification starts. Changing it halfway can mean paying for the same work twice.

Example: Money Train 4

Relax Gaming released Money Train 4 in September 2023 as the last game in its Money Train series. It was announced with more than twenty features and a 150,000x max win. Its main RTP is 96.1%, and one review says most UK casinos use a 94% alternative version. All those features need testing. Each RTP version is also its own math check. A game like this would likely land near the top of the cost range.

3. Testing Fees by Jurisdiction

Lab fees are only part of the bill. Regulators charge their own fees and each market has its own path. Michigan lists clear numbers for suppliers. The application fee is $2,500 and it is not refundable. The initial licence fee is $5,000 and the annual fee is $2,500. A Michigan memo also says the independent test lab tests the game and sends a certification letter. The board’s acceptance of that letter does not guarantee approval.

Example: Nolimit City

Evolution launched Nolimit City slots in six US states at the same time. Those were Connecticut, Delaware, Michigan, New Jersey, Pennsylvania, and West Virginia. Buffalo Hunter went first. Nolimit City games are described as certified by labs like GLI and iTech Labs. Newer releases include San Quentin Manhunt and Punk Rocker 3. A launch across six states means six sets of rules and paperwork even when the game stays the same.

4. Retesting and Recertification Costs

A certificate does not last forever. UK rules say minor updates can go out without outside testing, but major changes need a test house. If an RGS or RNG update touches many games, only a representative sample is retested. UK licensees also face an annual games testing audit. Guides list yearly monitoring at $2,000 to $5,000 and say re-certification is needed when paytables or rules change.

A few ways to keep the bill down:

• Sort each change as minor or major before you book a lab
• Group several updates into one retest when you can
• Use GLI-19 or GLI-33 certification when entering a new market since the lab can run a smaller delta test and not a full re-evaluation
• Keep the math model locked before you submit

How to Choose a Casino Game Testing Laboratory?

Pick a lab that your target regulators accept, and that can test your kind of game on your schedule. Start with the markets you want to launch in and check each regulator’s approved lab list. Then compare the lab’s accreditation and its testing coverage. Also look at how fast and helpful it is when you need a retest. Price matters, but fit matters more.

1. GLI, BMM and iTech Compared

The cost and time numbers come from industry guides and not from the labs so treat them as rough. GLI’s purchase of iTech did not end the iTech name. It still operates under its own brand and its clients can use GLI’s wider lab network. So the choice between them often comes down to which regulators know and accept each name. Ask each lab for a written quote and for a list of the markets where its reports are accepted.

2. Accreditation and Regulator Recognition

ISO/IEC 17025 is the main standard that shows a lab can produce valid results. Guides also point to ISO/IEC 17020 for inspection work and ISO 27001 for information security. Singapore requires its approved labs to keep ISO/IEC 17025. The UK list is built the same way. Many regulators will only accept a certificate from a lab on their own list, and those lists differ by market. Ontario goes a step further. 

The AGCO registers independent testing labs and does not test iGaming technology itself. In Michigan, the board accepts the lab’s letter but says that does not guarantee approval.

Ask these questions before you sign:

• Is the lab on the approved list for every market on your plan?
• Is its accreditation current and does it cover iGaming?
• Does it hold ISO 27001 for handling your source code?
• Will the regulator take the report as it is or ask for more?

3. Testing Coverage and Technical Expertise

Labs do not all cover the same ground. GLI says its RNG reviews can include source code review and the RNG period and range and the seeding process. eCOGRA works in four phases. They are a document review and a source code review, statistical testing and then reporting. The statistical step uses NIST SP 800-22 and the DIEHARD battery along with its own tests. 

Some labs also test whole platforms and not just games. BMM lists casino games and sports betting and lottery systems and platform testing. iTech lists RNG testing and game certification and platform audits. Pick a lab that has done your kind of product before.

Example: Lightning Roulette

Evolution’s Lightning Roulette launched in 2018 and mixes a live dealer with RNG multipliers. Each round strikes between one and five lucky numbers with payouts from 50x to 500x. XXXtreme Lightning Roulette later added Chain Lightning and Double Strikes that reach 2,000x. A game like this needs a lab that can test the live studio side and the RNG side together. 

In the UK, live dealer studios have their own standard called RTS 17. Before you hire a lab ask if it has tested a hybrid game like this.

4. Timelines and Retesting Support

Industry guides put most full certifications at about 4 to 12 weeks. A simple slot can be done in 2 to 4 weeks and a complex one takes 4 to 8. The biggest delays usually come from missing documents and not from the testing. A short pre-submission call with the lab can catch gaps early. GLI also offers pre-certification testing and regulatory advice. Retests matter too. In the UK a change to an RGS or RNG that affects many games means retesting only a sample. A lab that offers a smaller delta test for new markets can save you time later.

Questions to ask about support:

• Will you review my documents before the clock starts?
• How do you handle small updates and big ones?
• Can you retest a sample when a platform change hits many games?
• Who do I talk to when a test fails?

Example: Lightning Storm

Evolution’s Lightning Storm is a good picture of a hard test job. It has a 39-segment DigiWheel and five bonus games named Hot Spot, Monster Mash, Battery Charger, Fireball, and Lightning Storm. Evolution’s product chief said making it was like building six different games. A game like that will take a lab longer than a basic slot. It also means any later change to one bonus game could trigger a fresh round of testing. Ask your lab how it would scope that work before you start.

Build a RNG-Ready Casino Game with Idea Usher

Idea Usher builds casino games that are ready for lab testing from the first day of development. Our team brings 500,000 hours of coding experience and includes ex-MAANG and FAANG developers. We design the math and the RNG and the RTP and paytable together with the documents a lab will ask for. That way your game reaches certification with fewer surprises and less rework.

Custom Casino Game Math Development

Every game starts with a math model. We build it around your goals for RTP and volatility and hit frequency and max win. Then we turn it into the documents a lab reads first. A PAR sheet and probability tables and feature distributions are the usual set. The model has to be final before certification can start. Changing it halfway can mean paying for the same testing twice. So we lock the math early and simulate it before any code goes to a lab.

What you get from this step:

• A math model matched to your target market and player type
• A PAR sheet and probability tables in a format labs can read
• Simulation results that check the model against its own theory
• A clear record of every math version

Secure RNG Algorithm Integration

The RNG is the part labs look at the hardest. GLI asks for the final source code in full along with the compiled binaries. It then builds the files itself and takes a checksum to prove the code matches the tested build. We write and integrate the RNG with that in mind. The code is clean and commented and easy to trace from seed to outcome. We also keep the RNG description and version notes and signatures ready so the lab can follow the whole path without guessing.

Statistical testing is next in line. Labs run tests like chi-square across huge sets of outcomes. One public GLI report passed its test battery at the 95%, 98% and 99% confidence levels. We run our own checks before submission so a bias or a weak seed shows up on our side first.

RTP and Paytable Setup

Alternate RTP versions need care. Pragmatic Play certifies several RTP versions of its Sweet Bonanza games, and each casino picks which one to run. Each version is a separate math setup that needs its own check. The theoretical RTP also goes into the game report a UK lab files. So the number in your code and your paytable and your documents must all match. We set them up together so they do.

Testing Prep and Game Validation

Delays usually come from missing documents and not from the lab work itself. A short pre-submission review can catch gaps before the clock starts. We run that review for you. Our steps look like this:

• Check the code and math against the documents one more time
• Run functional tests on every payout rule and bonus trigger
• Package the source code, math files, rules and version records
• Support questions from the lab until the report comes back

Labs such as GLI, BMM, and iTech Labs each have their own submission style. Regulators also differ by market. We help you plan the lab choice and the target markets before the build starts so you do not pay to fix it later. Ready to build? Talk to the Idea Usher team and tell us about your game idea.

Conclusion

Casino game math and RNG certification decide if your game is fair and safe to launch. The math sets how often players win and how much the game pays back over time. The RNG makes sure every spin or card comes out randomly so nobody can guess it. A trusted testing lab checks both before the game goes live. Without that proof, most regulators will not give you a license and players will not trust your brand. So make math and certification a real part of your plan from day one. 

FAQs