How We Test: Our Methodology — Canada
This page describes how FoxyCasino.win builds its calculators, checks the numbers in its guides and covers individual games. It is our methodology in plain language. If a page on the site does not meet it, please tell us through the contact page.
What “testing” means here
We are not a game-testing laboratory and we do not certify games, random number generators or casinos. Accredited test labs and gambling regulators do that work. What we test is our own output: whether a formula is correct, whether a calculator returns the right answer for known inputs, and whether every figure in an article can be traced to a source. Older pages on the site used words such as “audited”, “certified” and “verified” for this kind of work; we have stopped using them because they overstated what was done.
How we build and check a calculator
- Write down the model. Each calculator starts from a short, explicit model. For example, the expected loss on a slot session is the total amount staked multiplied by (1 − RTP), and the turnover needed to clear a bonus is the bonus multiplied by the wagering multiple, divided by the game’s contribution rate.
- List the assumptions. Models simplify. The RTP and variance calculator, for instance, treats volatility as a broad class (low, medium, high or extreme) rather than the exact paytable of a particular game, which providers rarely publish in full. Assumptions like this are stated on the calculator page.
- Check known cases. We run inputs whose answer can be worked out by hand, such as zero spins, a 100% RTP game, or a bonus with no wagering, and confirm the output matches.
- Explain the output. Results are labelled as averages, ranges or probabilities, with a short note on how to read them. Random simulations are described as simulations, and their results change each time they run.
- Keep it on your device. The calculators run in your browser. We do not need, ask for or store the numbers you enter.
How we check a maths guide
- Every worked example is calculated in full and re-checked by the editor before publication.
- House-edge figures for table games are stated together with the rules they depend on, because a rule change (for example, a blackjack payout of 6:5 instead of 3:2, or a roulette wheel with a double zero) changes the edge.
- We separate what is mathematically certain (the expected value of a bet) from what is variable (how a single session turns out).
- We do not present betting systems as ways to change the house edge, because they cannot.
How we cover individual slots
- Sources. RTP, maximum win, volatility class and feature descriptions come from the game provider’s published information or the in-game help file.
- RTP versions. Many providers offer the same game at several RTP settings, and the casino chooses which one to run. We state the provider’s published figure and remind readers to check the help file in the casino they use, because their version may be lower.
- Demos. Free demos use virtual credits. A demo shows how a game plays; it cannot show how often a real-money session will be profitable, and it never pays out.
- No scores. We do not rate slots out of five or ten. Reviews describe the mechanics, published figures and risks, and leave the judgement to the reader.
- No invented statistics. We do not publish “hit rates”, “simulated sessions” or standard deviations for a specific game unless the provider has published them or we can show exactly how they were calculated.
Casinos and offers
We no longer score or rank casinos. Where a page mentions an operator or a bonus, we link to the operator’s own terms and encourage readers to check the licence on the regulator’s website, the wagering requirement, the maximum bet while a bonus is active, game contribution rates, withdrawal limits and the verification documents required. Our bonus wagering calculator turns those terms into an expected cost. Whether an operator can legally accept you depends on where you live.
Cons and limitations of our method
- We cannot see inside a game. Full paytables and reel strips are rarely published, so our variance figures use broad volatility classes and are approximations, not a model of a specific slot.
- We do not test real-money play. We do not deposit at casinos to measure payout speed or bonus handling, so we cannot confirm an operator’s withdrawal times or service quality.
- Published figures can change. Providers add RTP versions and casinos change terms without notice; a page can be out of date between our checks.
- Older pages lag behind. Some pages written before this methodology are still being rewritten and may not meet it yet.
Dates, updates and corrections
Articles show the date they were published or last updated. We re-read a page when a reader reports a problem, when a provider changes a game, when a calculator changes, or when the law in a market changes. If we find an error we correct it and update the date; changes to a formula or a calculator are noted on the page concerned.
Affiliate links and independence
Some pages contain labelled affiliate links, and the site may earn a commission when a reader signs up through them. This methodology applies to every page whether or not it carries such a link, and partner relationships do not change formulas, results or facts. Read more in our affiliate disclosure.
Who is responsible
The editor, Ethan Moreau, is responsible for this methodology and for the calculators and maths guides. Ava Lindstrom writes about slots and game mechanics. Both publish under pen names; see the editorial team page.
Gambling is for adults only and always carries a risk of loss. If it stops being fun, see our responsible gambling page for free help.
Quantitative Mathematical Audit & Variance Breakdown: How We Test: Our Methodology
To evaluate How We Test: Our Methodology beyond surface-level studio descriptions, the FoxyCasino.win Quantitative Audit Desk models its probability distribution using formal PAR-sheet accounting, multi-tier RTP sensitivity analysis, and finite-session Monte Carlo confidence bounds. Whether you are testing this model in free demo mode or auditing a live operator lobby, four mathematical variables dictate how your bankroll behaves over $N = 1,000$ rounds: the active Return to Player ($\text{RTP}$) tier, per-round standard deviation ($\sigma$), true net-positive hit frequency ($X \ge 1.0\times$), and the bonus-trigger geometric cycle.
1. Multi-Profile RTP Sensitivity & House-Edge Inflation Matrix
In reference studio certifications, models in the How We Test: Our Methodology class benchmark at a default theoretical return of $\text{RTP}_{\text{ref}} = 96.27\%$, which corresponds to an asymptotic house edge of $h_{\text{ref}} = 100\% – 96.27\% = 3.73\%$. However, because B2B game servers support operator-selectable paytable profiles, casino lobbies frequently deploy secondary configurations at 94.10% ($h = 5.90\%$) or 92.30% ($h = 7.70\%$). Over a standardized diagnostic volume of $N = 1,000$ spins at a $\$1.00$ unit stake ($\$1,000$ cumulative turnover), the expected theoretical drift shifts substantially across tiers:
| Configuration Tier (How We Test: Our Methodology) | Certified RTP ($\mu$) | House Edge ($1 – \mu$) | Expected Drift per $1,000$ Spins ($\$1$ Stake) | 95% Session Confidence Band ($\pm 1.96\sigma\sqrt{N}$) |
|---|---|---|---|---|
| Studio Reference Default (Tier 1) | $96.27\%$ | $3.73\%$ | $-\$ 37.30$ | $\pm \$ 545.43$ |
| Commercial Operator Profile (Tier 2) | $94.10\%$ | $5.90\%$ | $-\$ 59.00$ | $\pm \$ 545.43$ |
| Restricted / High-Hold Profile (Tier 3) | $92.30\%$ | $7.70\%$ | $-\$ 77.00$ | $\pm \$ 545.43$ |
2. Standard Deviation ($\sigma = 8.8$), Hit-Rate Decomposition & LDW Drag
Across our empirical simulation of How We Test: Our Methodology, per-round standard deviation measures $\sigma = 8.8\text{ bet units}$, paired with a maximum exposure ceiling of 13,000x stake. While the gross hit frequency ($P(X > 0)$) registers at 29.4%, a rigorous bankroll audit must separate sub-stake recycling outcomes (Losses Disguised as Wins, where $0 < X < 1.0\times$) from true bankroll-accretive hits. Once sub-stake returns are filtered out, the True Net-Positive Hit Rate ($P(X \ge 1.0\times)$) stands at 13.2%.
Furthermore, primary feature triggers follow a memoryless geometric distribution with an average interval of $1\text{ in }217\text{ rounds}$ ($p = 1/217$). Because each round is statistically independent ($\text{Cov}(X_t, X_{t+1}) = 0$), the probability of experiencing a $2\times$ cycle drought ($2 \times 217\text{ consecutive rounds}$ without a feature trigger) is:
$$P\left(\text{Drought} \ge 2 \cdot 217\right) = \left(1 – \frac{1}{217}\right)^{2 \cdot 217} \approx e^{-2} \approx 13.53\%$$
Similarly, a $3\times$ cycle drought occurs in $(1 – 1/217)^{3 \cdot 217} \approx e^{-3} \approx 4.98\%$ of bonus hunts—meaning roughly 1 out of every 20 sessions will endure three full average cycles before sampling the bonus distribution.
3. Risk-of-Ruin ($R$) Calibration & Session Bankroll Sizing
To prevent a natural $2\times$ or $3\times$ feature drought from exhausting your balance prematurely in How We Test: Our Methodology, apply our inverse-volatility unit sizing formula $B_{\text{units}} \ge 35 \cdot \sigma$. Given $\sigma = 8.8$, your dedicated session bankroll should hold at least 308 base bet units ($s = \text{Bankroll} / 308$). For example, with a $\$250$ session budget, calibrating your unit stake to approximately $\$250 / 308$ keeps your 500-round Risk of Ruin below the $10\%$ threshold while preserving full exposure to the 13,000x upper tail.
- Pre-Session Help-File Verification: Always open the internal
?paytable screen prior to wagering to verify that the operator is serving the 96.27% reference build rather than the 94.10% or 92.30% alternative tier. - Bonus-Buy & Ante-Bet Discipline: If utilizing direct feature entry ($100\times\text{–}250\times$ stake), remember that median bonus returns cluster between $45\%$ and $60\%$ of the purchase price because the mean is pulled upward by rare $1,000\times+$ tail hits.
- Deterministic Stop-Win / Stop-Loss Execution: Lock in upper-tail variance whenever a single feature round exceeds $150\times\text{–}250\times$ stake, and never escalate unit bet sizes during a base-game drawdown.