Intuition in games sounds appealing, but what stands behind real strategies tested not in words but in numbers? We spoke with Andrew Shepard, Head of Product at Roulette77 and author of the book A Gateway To The Roulette World. He has tested more than fifty systems, run hundreds of thousands of spins in simulators, and written dozens of analytical reviews. In this interview, Andrew explains in simple terms why thousands of rounds are necessary, why short sessions are misleading, and which testing methods deliver honest results. The conversation turned into a discussion about how theory becomes practice and why numbers are always stronger than feelings.
Andrew, in your book A Gateway To The Roulette World, you wrote that many beginners start with intuition. Why, in your opinion, are large data sets more important than a player’s “gut feeling”?
When I first started working with players, I really did hear the same thing: “I feel that red will come up soon” or “this system works, I tested it myself.” Intuition is, of course, fine, but in roulette it can play a cruel trick on you. Mathematics works regardless of whether we believe in it or not. One evening at the table does not reflect reality, because anything can happen in the short term. For example, red may come up ten times in a row, and you may think you have found a “gold mine.” But if we look at a million spins, we will see that red and black balance out and return to their mathematical probabilities. That is why it is more important to trust the numbers, not intuition.
How many spins are needed for a strategy to really show itself? Is a thousand enough, or are we talking about millions?
In my view, a thousand spins are more than enough to understand the strengths and weaknesses of a strategy. Let me explain why. In roulette, each bet follows strict mathematical probabilities, and even a thousand spins provide enough data to see how these probabilities manifest in practice. We have checked this many times: if a strategy breaks down, it becomes noticeable within the first few hundred spins. For example, the Martingale. In theory it looks flawless, but already in a series of a thousand spins, almost always a long losing streak appears, which leads to hitting the table limit or going bankrupt. This is visible right away; you do not need to wait for millions of attempts. Scaling this to millions of spins is pointless; you get the same picture, only stretched over time. That is why 1000 spins is the optimal balance between accuracy and practical value.
Describe your workflow: how exactly do you test a strategy from the initial idea to the final conclusion?
It all starts with an idea. Usually it is a system already circulating on forums or search engines. We take it and first break it down step by step: which bets, what logic, how the chip size grows or falls. In other words, we prepare a sketch, a draft model.
The next step is the simulator. Sometimes we write the model in JavaScript, sometimes we use Google Sheets. It depends on the complexity of the strategy. We set parameters: number of spins (usually hundreds of thousands), starting bankroll, betting limits. We run the simulation and get the “raw” data: how many wins, how many losses, where the betting peaks were.
After this comes the analysis. We build graphs, look at variance, average expectation, and the risk of complete bankruptcy. The final stage is the conclusion. We write a note or review with the results. And here is an important point: we never promise that a strategy will make money. We show how it behaves in the long run, what its strengths and weaknesses are. The player can decide for themselves whether to use it or not.
By the way, today readers have a tool that makes this work easier. You can describe the essence of a strategy and the rules for changing the bet to any AI assistant, and it will build the simulation itself. Most often, this turns into a small Python script that calculates the results and builds a graph. It is a quick way to test a system and see its dynamics, even if you have never written code before.
Let’s go into more detail about the analysis. Which parameters are most critical for you? Only return (RTP) or, for example, variance and risk of bankruptcy as well?
RTP is undoubtedly an important indicator. Roughly speaking, it shows what percentage of bets will be returned to the player in the long run. But the problem is that RTP does not explain how exactly you will get to that result. This is where variance and risk of bankruptcy come into play.
Imagine two scenarios. In the first, the strategy gives you many small wins and rare large losses. In the second, the opposite: you endure a series of small losses and then once in a hundred spins get a big win. From the point of view of RTP, both options may be identical. But the player’s experience will be completely different, and the risks as well. That is why in our tests we always look at three things:
- Variance — how much the result “swings.”
- Maximum drawdown — how much a player can lose before coming back into profit.
- Risk of bankruptcy — the probability that the bankroll will be wiped out under given limits.
It is the combination of these factors that gives a complete understanding. RTP answers the question of how much on average, while variance and risks explain how exactly. It is important for the player to know not only that they will face minus 2.7% in the long run, but also whether they are prepared for the path that will lead to that result.
You have already tested more than fifty systems. What has been the most surprising discovery in these experiments?
To be honest, the most surprising thing is how similar the results are to each other. More than once I have come across strategies presented as something extraordinary. But as soon as we run the tests, it turns out that they all eventually lead to the same thing: mathematics dominates, and the casino stays in profit.
Take, for example, the “Fisher system.” When I first heard about it, it seemed unusual to me — complex betting logic, many conditions. It seemed it might surprise. But after 500,000 spins, it turned out that the result was no different from simpler systems. Yes, the fluctuation graph is different, yes, the feeling of play is different, but the outcome is always the same long-term loss.
Many online resources simply copy strategies from each other. How does your approach differ from the “copy-paste” found on the internet?
The biggest difference is that we do not stop at theory. On the internet you can find hundreds of articles on strategies, and most of them really are copied from each other. There are lots of loud promises — “guaranteed win,” “casino secret revealed.” But the authors of these texts have never actually checked how these systems behave in practice. Our approach is different. We take a strategy and run it through simulators. Not 10 spins, not 100, but thousands. We count real numbers, build graphs, record drawdowns. As a result, we get a picture not based on nice words but on facts. In addition, we always specify the limitations. A strategy may work for short sessions or for players who are comfortable with high risk. But we honestly say that in the long run it will not save you. This approach requires more time and resources, but it provides value. Players do not get a copied forum article but a real analysis. That is what distinguishes us from “copy-paste” and makes the results useful.
This article was written in cooperation with Roulette77.