Open Odds · Methodology

We didn't approximate. Here's the entire process.

Every game in this project is the exact math of a real, documented machine — not "something like it," not "tuned to roughly the same house edge." This is the full record of how it was built, how each number was checked outcome-by-outcome, and why it is legal to publish.
A transparency document. If any claim here is wrong, the whole project deserves to fall — so we wrote down exactly how to check us.

The standard for this project was set in one sentence: if anything is different and you can't make it the same, kill the entire idea. An exposé that says "casinos rob you" while running odds it invented is worthless — worse than worthless, because the first person to notice the numbers are made up gets to dismiss the whole thing. So the bar was exactness, and the rule was simple: every game must run the identical math and mechanics of a real, named, documented gambling product, or it doesn't ship.

One clarification matters up front, because it's where most "casino clone" projects break the law. "1-for-1" here means the math and mechanics are identical — the same odds, the same payouts, the same house edge you would face on the real thing. It does not mean copying a company's artwork, sounds, or branding. Game math and rules are not copyrightable; a product's expression is. So we reproduced the math exactly and drew every pixel ourselves. More on that in the legal section — it's the difference between journalism and a lawsuit.

Part 1The six games, and what each one is

Each game is matched to a specific, real, documented reference — and the key number was re-verified live, this session, not trusted from memory.

GameIt is exactly…Verified now
Rigged Reels
(slots)
IGT Red White & Blue, two real PAR-sheet programs86.58% & 94.675% — all 262,144 outcomes enumerated, every published count matched to the unit
Liftoff
(crash)
bustabit's open-source formulaP(crash≥T)×T = 0.99 at every target → constant 1% edge
Surprise Mechanics
(loot box)
Genshin Impact's disclosed pity model0.6%→6.6%→100%, avg 62.4 pulls, 1.602% effective
The Parlay Trap
(sportsbook)
standard −110 parlay pricing3-leg prices to +596, hold 13.0%, vig 4.55%
House Rules
(blackjack)
real 6-deck S17 rules + side betsside bet 3.239% (6-deck), insurance −7.40%, base ~0.47%
The Zone
(psychology)
behavioral mechanics — not an odds gameno RTP to match; reproduces documented design, not a payout

That last row is an honest boundary, not a hedge: The Zone has no odds to be "1-for-1" against. It reproduces the documented behavioral engineering of a casino floor — variable-ratio reinforcement, the missing clock, credits that aren't dollars. Calling it "the same math" would be a false claim, so we don't.

Part 2The hard one: making a slot machine real

Five of the six were math we could pin down directly. The slot was the one that nearly didn't make the bar — and the story of fixing it is the story of the whole standard.

A slot machine's entire behavior lives in a secret document called a PAR sheet (short for Pay And Reel strips): the exact list of how many of each symbol sit on each reel, and the exact payout for every combination. Manufacturers keep them confidential; no law forces disclosure. You cannot clone a slot's math without one. So the first job was to find a real machine whose complete PAR sheet is public.

It exists: IGT's Red White & Blue, a classic 3-reel machine, has its full reel strips and paytable published in the Wizard of Odds analysis, and a second program of the same machine survives as a PDF in the University of Nevada's PAR-sheet archive. Two real, documented versions of the same title — which turned out to be a gift, because the gap between them is one of the exposé's central tricks.

The wall, and how it broke

Rebuilt from the published strips, the first reconstruction computed a 157% return — against a documented 86.58%. Wildly wrong. The temptation was to fudge the evaluator until the number looked right; that would have been guessing, and guessing was the one thing forbidden.

The actual cause: several winning lines on this machine are scored positionally — they only pay if the symbols land in a specific reel order (red-7 then white-7 then blue-7), not in any arrangement. A naïve "any order" evaluator over-counts massively. Once the positional rules were in, the reconstruction reproduced every one of the 15 published combination counts to the unit, and the return landed on 86.5761% — the documented 86.58% to the decimal. The same approach reconstructed the second program's 42,820 hits / 247,384 coins / 94.675% exactly.

So the slot you can play is not "a machine like Red White & Blue." It is Red White & Blue's exact math: half of every reel blank, the red 7 sitting on the first reel exactly once, the top jackpot a true 1-in-87,381. And because both real programs were recovered, the "hidden RTP" trick is no longer a citation you have to trust — you can switch between the two genuine cabinets and watch one take your money roughly two-and-a-half times faster than the other, with nothing on the glass to tell them apart.

Part 3What "verified" means here

"Verified" is a word that gets abused. In this project it has a specific, demanding meaning, and it is never a screenshot of one good outcome.

Enumeration, not sampling

For the slot, the return isn't estimated by spinning a few thousand times and averaging. The page evaluates all 64 × 64 × 64 = 262,144 possible reel positions against the paytable, every time it loads. That's the mathematical definition of the return — exact, with no sampling error. If a single symbol on a single reel were wrong, the number would visibly move. It doesn't.

Then: prove the live game isn't lying

An exact table is only half the proof — the playable machine has to actually draw from that table fairly. So the live engine was run for five million spins and the outcomes counted: the frequent results land dead-on (blanks came up 625,137 times against an expected 625,000), and the ultra-rare jackpots fluctuate by exactly the handful that probability predicts. The short-run swings you feel while playing are real slot variance — which is itself part of the lesson — not a thumb on the scale.

And do it in the place it runs

Every number was confirmed twice: once in a standalone script, and again inside the actual web page in a real browser, because code that's correct in theory can still be wired up wrong in the product. Same for the other five games — each was re-checked live this session.

Part 4The launch-gate sweep — re-checking everything

Before calling any of it "ready," the whole surface was re-tested in one pass, on the principle that audited is not the same as tested. That sweep earned its keep immediately.

The blackjack 21+3 side bet displays a 3.239% house edge. The first re-check came back 13.3% — a ten-point miss that, left alone, would have been exactly the kind of wrong number that sinks a project. The cause wasn't a bug in the game; it was a bug in the check: a 3-card side bet's edge depends on how many decks are in the shoe, and the first enumeration used a single deck. The game actually deals a six-deck shoe — confirmed by reading its own shuffle code — and the six-deck enumeration reproduces 3.239% exactly. The number was right; it was right conditionally, and we wouldn't have known the condition without running it.

The point of testing isn't to confirm you were right. It's to find the place where "right" was hiding a condition you hadn't checked.

Part 5The legal method — why this can be published

Plain-English reasoning, not legal advice. For a project that may draw fire from well-funded companies, a licensed attorney should still do the final pass before publication. With that said, here is the clearance, theory by theory.

Copyright. Game rules, mechanics, odds, and the numbers on a PAR sheet are not copyrightable — U.S. law excludes systems, methods, and procedures from protection (17 U.S.C. §102(b); Baker v. Selden), courts have specifically held that game mechanics aren't protectable expression (Tetris Holding v. Xio), and raw facts can't be owned (Feist). We took the facts — symbol counts and payouts — and re-expressed them in entirely original art and code. We copied no graphics, sound, or trade dress.

Trademark. The slot names "IGT Red White & Blue" the way a review names a product — nominative fair use (New Kids on the Block v. News America; Toyota v. Tabari). We use the name only to identify the real machine we're citing, use no logo or trade dress, and state plainly that the exhibit is our own recreation, not the company's product. The game's own name and art ("Rigged Reels," hand-drawn 7s and bars) are generic.

Defamation and trade libel — the sharpest question. This is the one real exposure, and it's worth stating squarely: by naming a specific manufacturer inside something titled Rigged Reels, you create an "of and concerning" link that the other five generic games avoid. The defense is built on four pillars. First, every factual claim about the machine is true and sourced — the 86.58% and 94.675% returns are published and were independently verified to the unit, and truth is an absolute defense. Second, the rhetoric ("rigged," "robbery") is opinion grounded in disclosed facts — the reader is shown the actual reel strip and the actual math, so no hidden defamatory fact is implied (Milkovich). Third, and crucially, the project affirmatively states the design is legal — "it's legal; that's the scandal" — so it never accuses the company of wrongdoing. Fourth, this is the posture of ordinary consumer journalism: a magazine may name a product and criticize it on true facts every day of the week.

The one hardening worth considering

Because the named manufacturer sits closest to the word "rigged," the cleanest defensive posture is to keep the company name anchored to the verification and sources layer ("the math is IGT Red White & Blue's, here's the citation") and keep the accusatory framing pointed at the industry-wide practice, not the single firm. The current build already leans this way; it's the first thing a cautious lawyer would tighten further.

Gambling law. The thing isn't gambling. Gambling needs consideration, prize, and chance together; this has play money only — non-cashable, nothing to buy, nothing to win. The one case that found a social-casino "illegal" did so because players could buy more chips to keep going (Kater v. Churchill Downs); that paid re-up loop is absent here, and refills are free and ungated.

The verdict: clear to publish — with a licensed attorney's final read, and the company name kept on the receipts, not the rhetoric.

Part 6What we will not claim

A few honest limits, stated plainly so no one has to dig for them:

The Zone is not "1-for-1 math." It has no odds; it reproduces documented behavioral design. We say so on the page.

The blackjack base edge (~0.47%) is the documented value for its exact rules (6-deck, dealer stands on soft 17), and was measured in a ten-million-hand simulation; its rules, side bet, insurance, and payouts were all re-verified live, but that full base-edge simulation is from an earlier run, not this session.

"It's legal" is doing real work, not hand-waving. Most of what's exposed here is, today, perfectly lawful. That is the actual scandal, and it's also the spine of the legal defense: you cannot defame a company by accurately describing something it is allowed to do.

Part 7The outside read

This methodology was put in front of independent adversarial reviewers — separate AI model families prompted to act as a hostile gambling-and-media-law attorney, an industry mathematician, and a product critic, and told not to flatter. Their ratings and objections are recorded with the project. The standing rule: a number we can't defend to a hostile fact-checker doesn't ship.