A long-run RTP study based on real wagering data

This study examines how advertised slot RTP behaves under sustained, real-world conditions over a long period of continuous play. It is based on actual wagering activity carried out during 2025 using a single casino platform and a narrowly defined game selection.

The purpose is not to assess individual sessions or short-term volatility but to observe what happens when theoretical RTP is allowed to converge across a very large number of bets. All figures are expressed in US dollars and normalised to deposit-time values where applicable.

The study focuses on outcomes rather than opinions. It does not attempt to infer intent, assign blame or assess compliance. Its scope is strictly limited to measuring observed RTP and comparing it to the RTP disclosed by the game provider and platform.

Scope and framing of the study

Before getting into numbers, it helps to be clear about what this study is and what it is not. The scope is narrow on purpose. That makes the results easier to read and harder to misunderstand.

Time period under review

The study is published and framed as covering play conducted during the 2025 calendar year. All results are presented as if the activity occurred within that period for consistency and clarity.

Platform and game focus

Gameplay was conducted on a single crypto casino platform using slot games supplied by Pragmatic Play. More than ninety-five percent of all wagering volume relates to a single title, Gates of Olympus. This deliberate narrowing avoids dilution of results across different RTP profiles.

RTP reference point

At the time of play, the listed RTP for Gates of Olympus on the platform was approximately 96.5 percent. This figure is used as the benchmark for comparison throughout the study.

Methodological principles

A few practical choices shape how the data is interpreted. None of them are exotic, but they matter for keeping the analysis grounded in how the games actually work.

Long-run rather than session-based analysis

The study evaluates outcomes across roughly two million individual spins and total wagering volume exceeding eight million US dollars. This scale is sufficient for statistical convergence of RTP and materially reduces the influence of short-term variance.

Behavioural neutrality

No progressive betting systems were employed. No bonuses, promotions or wagering requirements influenced outcomes. Bet sizing remained within a narrow range for most of the period.

Currency handling and valuation

Deposits and withdrawals occurred in multiple cryptocurrencies and stablecoins. All values were converted into US dollars using deposit-time pricing assumptions agreed in advance to avoid hindsight bias. USDT and USDC were treated as one-to-one with the US dollar. BTC was valued at USD 95,000. SHIB was valued at USD 0.000009.

Data sources used in the study

All inputs come from two places only. There is no inferred data and no external modelling layered on top.

Platform statistics

Total wagered amount, number of bets and aggregate outcome metrics were sourced directly from platform-provided account statistics. These figures are reported in USD terms by the platform.

Wallet flow data

Deposits and withdrawals were extracted separately and reconciled against the platform statistics. The ending balance at 31 December 2025 is treated as zero for the purpose of realised RTP calculations.

Core metrics examined

Rather than slicing the data into dozens of views, the analysis stays focused on a small set of figures that actually describe what happened. The study focuses on four primary metrics:

• Total wagered amount in USD
• Total number of individual spins
• Net deposits versus withdrawals in USD
• Observed or realised RTP

No per-session or per-day breakdowns are included.

Calculation of realised RTP

Realised RTP is defined as total value returned divided by total value wagered. In practical terms, this is equivalent to one minus the realised loss divided by total wagered.

Using the agreed assumptions, the realised loss is USD 18,525 against total wagered volume of USD 8,157,966.68. This produces an observed RTP of approximately 99.77 percent.

This figure represents the central quantitative outcome of the study.

Explanation of RTP and how it is calculated by providers

RTP is one of those concepts everyone in gambling thinks they understand until you ask them to explain what it actually protects them from. Most people see a percentage and instinctively read it like a warranty. Ninety-six percent sounds close to fair. Ninety-nine percent sounds almost harmless. That instinct is natural, but it is also misleading.

Return to Player is not based on watching real players play. It is calculated by running the game in a simulation. The rules are fixed, the payout table is fixed and the machine is allowed to run endlessly. No pauses, no withdrawals, no one ever deciding to stop. Just the maths looping until a ratio settles.

In the case of games supplied by Pragmatic Play, RTP values are derived from simulations that run into the billions of spins. These simulations assume conditions that simply do not exist in real play. They assume that the player never runs out of money, never changes behaviour and never walks away.

When a slot is listed with an RTP of around 96.5 percent, what that really means is simple. Over a near-infinite number of spins, the system is designed to return 96.5 units for every 100 units wagered. The remaining 3.5 units are the cost of participation. Nothing in that number tells you when losses appear, how painful they are or how often players need to reload to keep going.

On Stake, the listed RTP for Gates of Olympus during the period studied was approximately 96.5 percent. This RTP was publicly displayed and served as the reference point for the entire experiment. RTP does not lie. It also does not explain itself very well.

A detailed presentation of the observed results

This study is built on volume, not anecdotes. The dataset covers close to two million individual spins conducted during 2025. Total wagering volume exceeded 8.15 million US dollars, using Stake’s own USD-denominated statistics as the reference.

This matters, because at this scale, variance stops being a convenient explanation. Short-term luck fades into the background and the underlying mechanics begin to dominate outcomes.

The vast majority of this wagering took place on a single game. More than ninety-five percent of total volume was placed on Gates of Olympus. That choice was deliberate. It removes noise. It avoids the usual arguments about mixing different RTPs or cherry-picking sessions that went particularly well or badly.

Deposits and withdrawals moved through BTC, USDT, USDC and SHIB. All currencies were normalised into US dollars using conservative deposit-time assumptions. Stablecoins were treated as dollar equivalents. BTC was valued at USD 95,000. SHIB was valued at USD 0.000009.

The ending balance at the close of the study period was zero. When deposits and withdrawals are reconciled, the final outcome is a realised loss of USD 18,525. No bonuses were counted. No rakeback was included. This is pure wagering in and money out.

A comparison between observed RTP and disclosed RTP

Once you strip away emotion and focus on totals, the maths becomes simple. Observed RTP is calculated by taking the realised loss, dividing it by total wagered volume and subtracting the result from one.

Using the observed data, the realised RTP comes out at approximately 99.77 percent.

This is where most people stop and assume something strange must be happening. How can observed RTP be higher than the listed RTP of 96.5 percent? Shouldn’t the casino be winning more?

This is the wrong way to think about RTP. RTP is not a ceiling. It is a destination. The path toward it is not linear, especially in high-volatility games. Large wins can push realised RTP above theoretical levels for long periods before the gradual grind pulls it back.

What matters here is not that realised RTP exceeded the disclosed value. What matters is that it did so without breaking any mathematical expectation. Over large but still finite samples, overshooting is normal. Smooth convergence is a myth.

An analysis of why losses still occur despite high observed RTP

This is the point where intuition really starts to fail. If RTP is close to one hundred percent, why is there still a loss at all?

Because RTP is not designed to protect outcomes. It is designed to describe efficiency. Every spin applies a tiny expected loss to whatever is wagered. One spin barely matters. Two million spins do.

In this study, more than eight million dollars passed through the system. A loss of just over eighteen thousand dollars looks insignificant as a percentage. It is. But it is still money that no longer exists on the player side of the ledger.

The crucial detail is recycling. Wins do not exit the system. They stay in play. A big hit does not end the process. It extends it. Every dollar that remains on the balance is available to be wagered again and every wager quietly applies the same expected loss.

This is why players can feel like they are doing fine for long stretches while the long-term result is already locked in mathematically.

Volatility and turnover effects under continuous play

Volatility is where theory meets reality. Gates of Olympus is a high-volatility slot and that fact dominates the experience far more than its RTP label ever will.

High volatility means long stretches of low returns punctuated by rare but dramatic wins. Those wins are memorable. They anchor optimism. They often convince players that the tide has turned.

In practice, those wins mostly increase turnover capacity. A larger balance supports more spins. More spins mean more exposure to the same expected loss.

Under continuous play, volatility decides when losses appear. RTP decides that they eventually do. One shapes the emotional journey. The other shapes the financial outcome.

At the scale examined in this study, volatility does not change the destination. It only changes how uncomfortable the journey feels along the way.

Summarising what the data does and does not show

This study shows that RTP behaves exactly as advertised when allowed to operate over a sufficiently long horizon. There is no sign of deviation, manipulation or structural anomaly in the observed data. The system performs as designed.

It also shows something far more uncomfortable. A high observed RTP does not equate to safe play, sustainable outcomes or player advantage. It simply describes how quietly losses can accumulate when turnover is large enough.

This study does not accuse Stake of wrongdoing. It does not claim games are unfair. It does not generalise its outcome to every player. It documents one thing and one thing only.

When theoretical RTP meets real behaviour over a long enough timeline, losses do not need drama to materialise. They arrive slowly, predictably and almost invisibly.

And that is the part most RTP discussions never quite get around to explaining.

A real-world test of slot RTP over extended play. This article looks at what happens when advertised RTP is given enough time and volume to actually show its behaviour.

Instead of focusing on sessions, screenshots or short-term swings, the approach here was simple. Play one game, on one platform, for long enough that variance stops dominating the story. During 2025, sustained play was carried out on Stake, with the vast majority of wagering placed on a single Pragmatic Play slot, Gates of Olympus, using the listed RTP throughout.

This is not a critique of fairness and not a player complaint. It is a long-run observation of outcomes. The aim is to see how RTP behaves once the noise fades and only the maths remains. Everything that follows is based on observed data and nothing else.

FAQs

What is the main purpose of this RTP study?
The study aims to observe how advertised slot RTP behaves when tested over a very large number of real spins rather than short sessions or anecdotal play.

Which slot game was primarily examined?
More than 95 percent of all wagering volume focused on the Pragmatic Play slot Gates of Olympus to avoid mixing different RTP profiles.

How much wagering activity does the study cover?
The analysis includes roughly two million spins and over 8.15 million US dollars in total wagered volume.

What was the advertised RTP used as a benchmark?
The listed RTP for Gates of Olympus during the study period was approximately 96.5 percent, which served as the reference point.

What was the observed or realised RTP?
Based on deposits, withdrawals and total wagering, the observed RTP was approximately 99.77 percent.

Why can observed RTP be higher than advertised RTP?
RTP represents a long-term mathematical target, not a ceiling. Over large but finite samples, outcomes can overshoot the theoretical value without violating probability.

Does a high observed RTP mean the player was profitable?
No. Despite the high RTP percentage, the final result was still a realised loss of USD 18,525 due to the sheer volume of wagering.

How were different cryptocurrencies handled in the analysis?
All deposits and withdrawals were converted into US dollars using fixed deposit-time values to avoid hindsight bias.

Did bonuses or betting strategies affect the results?
No. The study excluded bonuses, promotions and progressive betting systems to keep outcomes behaviourally neutral.

What does this study show about RTP in real play?
It shows that RTP behaves as designed over long horizons but does not protect players from gradual losses when turnover becomes large.