Know every tool.
Understand every result.

An ordered rule that reacts to a win, loss, streak, bet count or another threshold. Rules can reset, set, add to or change a bet or win chance, change direction, or stop play.

Add all observed values, then divide by the number of observations. Large wins or losses can move the average sharply.

Example: Profits +20, +20 and −100 HD: mean = −20 HD.

The average of each simulated session’s worst drop from its own earlier balance peak.

A session that stops because the next required bet is larger than its remaining play balance. It does not necessarily mean its balance reached exactly zero.

The starting virtual HD Coin balance available to fund a session or simulation.

The share of tested sessions that did not stop for insufficient balance or invalid next settings. It is not the share that finished with a profit.

The initial bet amount chosen for a session; reset-to-base rules return to this amount.

How large the strategy’s high-end bet amounts became compared with its starting bankroll. Strategy DNA uses the 90th-percentile largest bet.

The current configured percentage target for a winning Dice roll. Advanced Play can change it during a session.

The order Advanced Play rules are evaluated. Two strategies with the same rules in a different order can produce different bet sequences.

An estimate of the upper end of a risk rate supported by the sample and its calculation assumptions. It is not a guarantee or an absolute maximum.

A plain-text table format that can be opened in spreadsheet programs. A CSV export contains the rows included by that particular tool, not necessarily all account history.

A simulation whose outcome repeats when the seed, strategy, settings and engine rules are kept identical.

The fall from an earlier high balance to a later lower balance during a session. Maximum drawdown is the largest such drop.

Example: From a peak of 150 HD down to 120 HD is a 30 HD drawdown.

The bankroll remaining when the session ends.

The number of simulated sessions stopped by insufficient balance divided by the total sessions tested.

Whether a session continued long enough to encounter every forced outcome planned by Stress Test Lab. Early-stopped sessions may have only partial exposure.

A one-time trigger for the first qualifying run of N consecutive wins or losses in a streak cycle; its adjustment affects subsequent play.

New roll sequences used to retest Optimizer candidates after they were selected. This helps reveal candidates that looked good due to random luck in the initial batch.

Virtual units used on HD4K Play. HD is the short label in dense research charts. HD Coins have no cash value and cannot be redeemed for money or prizes.

A color-coded grid where each square represents a tested combination of two variables. The colors compare the selected metric within the tested sample.

The mathematical advantage built into a game’s payout model. HD Dice research uses a nominal return multiplier of 99 divided by configured win chance, equivalent to a nominal 1% edge before discrete-roll and rounding effects.

A run that ends because a next bet or chance would violate the simulation’s allowed limits; it is tracked separately from running out of usable balance.

A shared strategy snapshot that can be opened using its exact link but does not appear in the searchable public community directory.

The greatest number of losses in a row observed within a session or across a collection of tested sessions, depending on the displayed label.

The share of simulated sessions that reached the configured maximum number of plays, rather than ending earlier for a target, stop-loss or another reason.

A per-session limit on the number of simulated rolls attempted. A session can end earlier due to targets, stop rules, insufficient balance or invalid settings.

The middle outcome after results are sorted. Roughly half of the observed outcomes lie on either side. It can differ sharply from the average.

Many independent virtual sessions run under the same rules to examine the spread of possible outcomes instead of relying on a single run.

Ending balance minus starting balance; negative values mean a loss of HD Coins in that session.

Example: Start 100 HD, end 125 HD → +25 HD net profit.

Recorded net profit divided by total amount wagered, expressed as a percentage. It is not the same as a win rate.

A chart or table showing how many tested sessions fell into different profit, drawdown or length ranges.

A lower-end sample percentile: about 10% of observed values are at or below this result. It is not the worst possible outcome.

The middle value in sorted results (median).

The 90th-percentile threshold: about 10% of observed values are at or above it. For drawdown, larger values indicate greater downside pressure.

A factor that converts a winning stake into the payout including the returned stake. Nominal Dice multiplier = 99 ÷ win chance (%) before discrete-roll/rounding effects.

Example: 49.5% chance gives a nominal 2× payout multiplier.

The highest balance reached so far within a session. Drops from prior peaks are used to measure drawdown.

The share of sessions that ended with net profit above zero. This is different from the share of individual bets that won.

A user-set positive net-profit threshold at which a research session stops when reached. A target-hit rate only exists when a target is enabled.

An intentionally published strategy snapshot that can appear in the Community Strategy Library and be inspected by visitors.

A return to a prior balance peak after a decline. Average recovery bets counts only dips that did recover; it does not imply all losses were recovered.

The difficulty of getting back to an earlier peak. Strategy DNA displays the percentage of sessions ending with an unrecovered drop.

A text value used to generate repeatable pseudo-random simulation streams. It is not a prediction of live future rolls or a live game seed.

In these HD4K simulations, an estimate based on how often sessions cannot afford the next required bet within the tested time horizon—not a lifetime mathematical certainty.

A Dice win condition: win if the roll is strictly above the Over target or strictly below the Under target. Equality is not a win in the research engine.

Makes your own editable Vault strategy from a published snapshot. The creator’s original and your copy have separate ownership and version history.

A sequence of virtual bets run under chosen settings until an exit condition occurs or the maximum play count is reached.

How many bets were actually completed before a session stopped. A maximum-plays setting is an upper limit, not a promise that every session uses all plays.

The percentage of sessions not terminated by insufficient balance or invalid play settings. A surviving session can still lose HD Coins or stop at a stop-loss.

How many virtual sessions were tested. Larger samples usually make estimated rates less noisy, but cannot guarantee outcomes or remove model limitations.

A configured negative net-profit threshold that ends a session after reaching the selected loss amount. A stop-loss is not a limit on possible one-bet slippage or a guarantee.

The number of simulated sessions that ended by reaching the enabled stop-loss, divided by all sessions tested.

A copy of strategy settings at the moment of publishing or saving a test. Later edits to the original do not silently change it.

A stored restore point for the saved Advanced Play rules. Version history lets you inspect or restore earlier definitions.

Survival asks whether the strategy kept enough balance and valid settings to avoid a forced stop; profitability asks whether ending balance exceeded starting balance. They are not interchangeable.

Number of sessions that ended because the enabled profit target was reached divided by the number of tested sessions.

The sum of all amounts staked across the bets included in the record or simulation. It is not necessarily money lost.

A standardized set of simulation settings used to compare public community strategies; its results are sample-specific rather than proof of performance.

How widely results vary across tested sessions. Strategy DNA shows the P90–P10 profit range relative to starting bankroll as one spread measurement.

Winning bets divided by total bets. It is not the same as percentage of sessions that end profitably.