Price and return
| TERM | DEFINITION | WHERE THE JOURNAL USES IT |
|---|---|---|
| Return to player (RTP) | The average fraction of every stake a game pays back over the long run, computed from its rules and payouts. A 97.22% return means the house keeps 2.78 of every 100 units staked, on average. | Sic Bo, every box priced |
| House edge | One minus the return: the share of stakes the house keeps on average. It is a property of the game, not of the player’s luck or history. | Sic Bo, roulette keys |
| Expected value (EV) | The probability-weighted average of all outcomes of a bet. A negative expected value means the bet loses money on average, whatever any single result does. | Kelly criterion |
| Total-return multiplier | A payout expressed as stake-back-plus-win, so ×2 on an even-money bet returns the stake and an equal win. Most tables on this site use this convention. | Roulette keys, Sic Bo |
| Paytable | The published mapping from outcome to payout. On this site paytables are read from engine source and the return is computed cell by cell rather than quoted. | Keno risk levels, Tairyō |
| Hit rate | The share of rounds that pay anything at all. It is independent of return: a game can pay often and little or rarely and much at the same price. | Bingo, Fukubukuro |
| Volatility | How widely individual results spread around the average, usually measured as the standard deviation of the payout in stakes. A “risk level” changes volatility while leaving return almost untouched. | Keno risk levels |
| Variance | The square of the standard deviation; the technical measure behind volatility. Higher variance means longer swings for the same expected value. | Risk of ruin |
Bankroll
| TERM | DEFINITION | WHERE THE JOURNAL USES IT |
|---|---|---|
| Risk of ruin | The probability that a bankroll reaches zero before a target, given the bet size and the game. With a negative edge and no stopping rule it is one. | Risk of ruin, Archie Karas |
| Gambler’s ruin | The theorem behind risk of ruin: a finite bankroll against an opponent with more money and any edge loses everything with probability one, given enough rounds. | Archie Karas |
| Kelly criterion | A bet-sizing formula that maximises long-run growth: stake the fraction of bankroll equal to edge divided by odds. With a negative edge it returns zero. | Kelly criterion |
| Stopping rule | A decision, made before play, about when a session ends: a loss limit, a win target or a time. It is the only lever that changes the chance of ruin without changing the game. | Responsible play toolkit |
| Bankroll | The money set aside for play, as distinct from money for anything else. Bet size is meaningful only as a fraction of it. | Risk of ruin |
| Gambler’s fallacy | The belief that a run of one outcome makes the opposite outcome due. On independent events the next result carries no memory of the last. | Gambler’s fallacy, Monte Carlo 1913 |
| Hot hand | The belief that a run of successes makes the next success more likely. Real in some skill processes; impossible on a seeded game, where rounds share nothing. | The hot hand |
| House money effect | The tendency to bet more freely with money recently won, as if it were not yet yours. The wallet keeps one balance; the effect lives in the player. | The house money effect |
Provably fair
| TERM | DEFINITION | WHERE THE JOURNAL USES IT |
|---|---|---|
| Provably fair | A scheme in which the house commits to its randomness before play and reveals it after, so any round can be recomputed by the player. It proves the randomness was fixed and honest; it does not remove the house edge. | What a hash commitment proves |
| Server seed | The house’s secret random string for a period of play. Its hash is shown before you bet; the seed itself is revealed when the pair is retired. | Verify a casino round |
| Client seed | A string you can set yourself, mixed into every round. It cannot change your odds; it guarantees the house could not have chosen a server seed to fit your input. | Illusion of control |
| Hash commitment | Publishing the SHA-256 fingerprint of a secret so it can be checked later. Two different seeds cannot share a fingerprint, so the revealed seed must be the committed one. | What a hash commitment proves |
| HMAC-SHA256 | A keyed hash: the server seed is the key and the message names the round. The output is the round’s raw randomness, identical for anyone with the same inputs. | Verify a casino round |
| Nonce | A counter that goes up by one with every bet across all games on the site, so no two rounds share inputs. | What the cursor counts |
| Cursor | A counter inside a round: cursor 0 is the first random number the game needs, cursor 1 the second. A bingo round reads 63; roulette reads one. | What the cursor counts |
| Seed pair | The server seed and client seed in force together. A pair is “rotated” when the server seed is revealed and a new one committed. | Illusion of control |
| Verifier | The code, in your browser, that recomputes a round from the revealed inputs. On this site it shares its functions with the demo engine, so the two cannot drift. | One rounding, not many |
| Uniform number | The value between 0 and 1 that a hash is turned into before a game maps it to a pocket, card or item. How the bytes become that number must match the server bit for bit. | One rounding, not many |
| Randomness beacon | A public source of random numbers, such as drand, published on a schedule and verifiable by anyone. An alternative to a house-held seed for some games. | Inside drand |
| VRF | A verifiable random function: a random output with a proof that it was computed correctly from a known input. One of four ways to prove a random number. | Four ways to prove a random number |
Games and crypto
| TERM | DEFINITION | WHERE THE JOURNAL USES IT |
|---|---|---|
| Infinite shoe | A card game in which every card is an independent draw from a full deck, so the same card can appear twice and removal never changes the odds. Counting has nothing to count. | Dealer bust odds, MIT team |
| Natural | A two-card 21 in blackjack. Paid 3:2 here; the 6:5 alternative costs about 1.42% of every stake. | 3:2 or 6:5 |
| Sector targeting | Predicting where a physical roulette ball will land from its speed and the wheel’s, before betting closes. Impossible on a hashed wheel, which has no ball in motion. | The Ritz laser scanner |
| Wheel bias | A physical roulette wheel whose pockets are not equally likely, found by recording thousands of spins. A defect, not a strategy. | Wheel bias |
| Confirmation | A block added to a blockchain on top of the one holding your transaction. Each one makes reversal harder; six is the customary threshold for value at rest. | What a confirmation is |
| Network fee | The price a blockchain charges to include a transaction, set by size or computation, not by amount. It is why dust exists and why withdrawals have minimums. | Crypto fees, Dust |
| Dust | A balance too small to move because the network fee would consume it. Not lost; it can be staked or topped up. | Dust |
| Destination tag | A second identifier some coins require alongside the address, naming the recipient inside a shared wallet. Without it the funds arrive at the address but not at you. | What the withdraw window checks |
Definitions are written for this site’s engines and may differ in detail from other operators’ usage. Every game figure quoted on the linked pages is read from engine source; this page carries no figures of its own.
FAQ
What is the difference between RTP and house edge?
They are the same number from two sides. RTP is the share of stakes paid back on average; the house edge is one minus RTP. A 97.3% RTP is a 2.7% house edge.
What does provably fair actually prove?
That the house committed to its randomness before you bet, that the revealed seed is the committed one, and that every result follows from the seed pair and round number. It does not prove the game has no house edge.
What is the difference between a nonce and a cursor?
The nonce counts rounds across the whole site; the cursor counts random draws inside one round. Together with the seed pair they identify every number the engine produces.
What is volatility in a casino game?
How widely results spread around the average, usually the standard deviation of the payout in stakes. Risk levels change volatility while leaving the return almost unchanged.
What is risk of ruin?
The probability a bankroll hits zero before a target. It depends on bet size relative to bankroll and on variance; with a negative edge and no stopping rule it is one.
SOURCES & REFERENCES
- The Betkyo Journal articles linked in each row, all of which read their figures from engine source
THE GAMES IN THIS ARTICLE
