Vault Breakers: Super Score — Design Specification
Date: 2026-08-19
Status: approved design; ready for implementation planning
Stable Game ID: vault-breakers-super-score
Target RTP after tuning: 96.00%
Lifecycle: concept/design only; no canonical config, runtime build or evidence exists
1. Purpose
Create an original high-volatility 6×5 tumble slot about a modern bank robbery. The gameplay loop uses publicly described mechanics from Gates of Olympus Super Scatter as an initial reference baseline, while all branding, characters, symbol identities, presentation and final mathematical tuning belong to Vault Breakers.
The reference baseline is not a claim that Pragmatic Play's complete math has been reproduced. Only publicly confirmed rules and values are carried into the first draft. Undisclosed paytable, weights and probabilities are not guessed.
2. Research boundary
Official Pragmatic Play sources confirm:
- wins for 8–30 matching symbols anywhere;
- tumbles;
- multiplier symbols from 2× to 500× whose values add together;
- 4+ Scatter symbols award 15 Free Spins;
- Free Spins can retrigger;
- Super Scatter awards of 100×, 500×, 5,000× and 50,000×;
- a 50,000× maximum win;
- two Bonus Buy modes in eligible markets, with Super Free Spins multipliers starting at 10×.
Sources:
The original game's exact symbol paytable, strip/refill distribution, multiplier weights, Scatter probabilities, Bonus Buy EV and RTP composition are outside this confirmed source set. They are not copied into this design.
3. Chosen approach
Selected: reference-faithful unbalanced baseline
Preserve the publicly confirmed mechanic and values for the first playable draft, but use an original theme and symbol set. Treat that draft as an instrumented baseline, then tune the complete game to 96.00% RTP.
Rejected alternatives
- Tune immediately. Rejected because the engine lacks several required mechanics and a tuning run before runtime parity would optimize a model the player path cannot reproduce.
- Attempt an exact clone. Rejected because the complete source math is not public and the product must not copy Pragmatic Play's branding, art or unverifiable probability model.
4. Theme and identity
High concept
The player joins a Mastermind-led crew breaking into a modern high-security bank. Winning loot is removed from the vault grid and replaced by new falling items. The Mastermind can inject hacking/explosive multiplier charges into a winning sequence. Scatter keys open the main vault; rare Super Scatters award instant scores while opening it.
Main character
The Mastermind is an original woman hacker positioned beside the grid. She is a presentation character only: animation never decides whether a multiplier lands or what value it carries.
Symbol inventory
The four multiplier appearances share one mathematical role and value pool. They are not four separate pay symbols.
5. Base game
Tumble sequence
- Generate the initial 6×5 grid.
- Evaluate every regular symbol independently by its visible count.
- Award at most one count-band result per regular symbol for the current grid.
- Remove every regular symbol that participated in a win.
- Drop surviving symbols downward within each reel.
- Fill empty cells with server-generated symbols.
- Repeat until the current grid has no regular-symbol win.
- Settle one spin result after feature and multiplier processing.
Scatter-class symbols do not form ordinary wins. A Scatter that is not part of a feature entry survives like any other non-winning symbol: it falls with its column and can coexist with newly refilled Scatters on a later snapshot. A multiplier charge is a one-shot event rather than a persistent blocker: its value is recorded once, its cell becomes empty, and that cell participates in the same deterministic drop/refill step as removed winning symbols. Simulation and player runtime must use this identical rule.
6. Multiplier charges
Base game
- A multiplier charge may appear on the initial grid or a later tumble grid.
- Its value is generated by the server from a weighted pool bounded by 2× and 500×.
- Each charge is collected at most once, then removed before the next tumble snapshot.
- Values of all eligible charges in one paid spin's tumble sequence are added.
- The sum multiplies the aggregate regular-symbol win of that sequence.
- A sequence with no regular-symbol win receives no multiplier award.
- The multiplier accumulator resets after the paid spin settles.
Free Spins
- Each Free Spin has its own independent tumble sequence and multiplier accumulator.
- Eligible charges from that Free Spin are added and applied to that Free Spin's aggregate regular-symbol win.
- The accumulator resets before the next Free Spin.
- It does not persist across all 15 Free Spins.
The initial reference pool permits values from 2× through 500×. The exact value set and weights are math inputs, not inferred from the presentation assets.
7. Scatter and Super Scatter
Snapshot rule
Scatter triggers are evaluated on one current 6×5 grid snapshot. Scatter symbols observed on different tumble snapshots are not accumulated together. Either the initial grid or any later tumble grid may independently satisfy the trigger.
Initial Free Spins trigger
SCATTERandSUPER_SCATTERboth count as scatter-class.- 4+ scatter-class symbols on one base-game snapshot award 15 Free Spins.
- One paid spin can award this Free Spins entry at most once; later snapshots cannot open a second sequence from the same wager.
- The current paid tumble sequence completes according to its authoritative outcome before the Free Spins presentation begins.
Super Scatter instant prize
An instant prize is payable only when the same base-game snapshot also satisfies the 4+ scatter-class Free Spins trigger.
SUPER_SCATTER without a qualifying 4+ scatter-class trigger pays no instant prize. A 4+
Super Scatter result reaches the global 50,000× cap; the server settles the game cycle and does
not generate additional Free Spin outcomes.
8. Free Spins
- Initial award: 15 Free Spins.
- Every Free Spin uses the same 6×5 count-anywhere/tumble loop.
- 3+ scatter-class symbols on one Free Spin snapshot add 5 Free Spins.
- Retriggers are repeatable; no arbitrary count limit is imposed below the global safety bound required to guarantee termination.
SUPER_SCATTERcounts toward retrigger but pays no additional instant prize inside Free Spins.- Reaching the global 50,000× cap ends the game cycle immediately and deterministically.
9. Bonus Buy design targets
Availability is market-controlled. These prices are reference baseline values and must be recomputed against the final 96.00% model.
There is no player choice after the purchase commits. A buy is one wagered operation with its own idempotency identity and one eventual settlement.
10. Runtime state and replay
The server is authoritative for grids, tumbles, wins, multiplier charges, Scatter classification, retrigger awards and settlement.
Durable Free Spins state must include:
- parent paid round ID and sequence ID;
- mode (
standardorsuper); - spins remaining and current Free Spin index;
- current tumble index and current grid when a step is open;
- accumulated win for the complete feature;
- multiplier total for the current Free Spin;
- awarded retriggers;
- cap state, expected next step and settlement status.
Replay with the same Idempotency-Key returns the same player-visible result. It must not mint a new grid, multiplier value, retrigger, debit or settlement. Reconnect resumes the unresolved step before enabling a new wager.
11. Mathematical lifecycle
Phase A: reference baseline
Implement only publicly confirmed values:
- 6×5 grid and 8+ count-anywhere threshold;
- 2×–500× multiplier bounds;
- 4+ Scatter for 15 Free Spins;
- 3+ Scatter retrigger for +5 Free Spins;
- Super Scatter table 100×/500×/5,000×/50,000×;
- 50,000× cap;
- 100×/500× Bonus Buy design prices and 10× minimum in Super Free Spins.
The following remain unset until an authored baseline and tuning record exist: regular-symbol paytable, refill weights, multiplier value weights, Scatter and Super Scatter frequencies, Bonus Buy EV, hit rate and volatility. An unset value blocks publishing; it is not filled with a guess.
Phase B: tune to 96.00%
The final report separates:
Exact results and sampled estimates remain in separate tables. Seeded Monte Carlo evidence must record sample count, seed scheme, runner identity, backend Git revision, config hash and confidence interval. Internal simulation is not certification.
12. Backend architecture gate
The existing Math Studio gates_of_olympus_style pack is a template, not a canonical game and
not an implementation base that can be renamed. It currently differs from this design:
waysevaluation is consecutive-reel ways, not count-anywhere;layout_multipliersupplies a fixed multiplier rather than weighted charges;tumblesupplies a progressive step multiplier rather than charge accumulation;- Free Spins retrigger awards the initial spin count, not a separate +5;
- Super Scatter and its instant-prize table are absent;
- Standard/Super Free Spins pools are absent;
- Buy options are not a complete player-runtime capability.
Implementation must add reusable, explicitly reported capabilities before creating the game pack:
- count-anywhere evaluator with deterministic winning-cell reporting;
- weighted multiplier-charge generation and per-sequence aggregation;
- a retrigger award distinct from the initial Free Spins award;
- scatter-class aliases plus conditional Super Scatter instant prizes;
- Standard/Super Free Spins multiplier pools;
- durable Bonus Buy execution if either buy is exposed to players.
Simulation-only support is insufficient. If the player runtime cannot execute a declared mechanic, capability reporting and publication must fail closed.
13. Error and cap behavior
- Invalid or incomplete math config is rejected before publication.
- Unsupported mechanics are reported as unsupported, not silently skipped.
- Math failure before persistence produces no game round; the paid path follows the existing exact refund/transaction contract.
- A failure after durable commitment is reconciled through the same operation; no second wager.
- The 50,000× cap applies once to the aggregate game-cycle award, including instant prizes, base wins, Free Spins and multipliers.
- A cap hit terminates further mathematical generation, not merely presentation.
14. Verification gates
Engine capability tests
- count-anywhere boundaries at 7, 8, every pay band and 30;
- simultaneous wins of multiple regular symbols;
- winning-cell removal and deterministic refill order;
- multiplier charges on initial, intermediate and losing-final snapshots;
- multiple charge aggregation and reset between spins;
- 4+ mixed Scatter/Super Scatter triggers;
- Super Scatter table and non-trigger non-payment;
- 3+ FS retrigger awards exactly +5;
- Standard versus Super multiplier-pool constraints;
- cap application and early termination.
Money and recovery tests
- exactly one debit per paid base spin or Bonus Buy;
- no debit for tumbles or Free Spin steps;
- identical replay for the same Idempotency-Key;
- reconnect resumes the same grid/tumble/multiplier state;
- one settlement despite retries and lost races;
- cap hit cannot generate or credit later Free Spins.
Math evidence
- deterministic golden vectors and reference/runtime parity;
- exact checks wherever tractable;
- seeded baseline simulation;
- tuning evidence reconciling to 96.00% within an explicitly approved tolerance;
- max-win reachability evidence;
- separate Standard and Super Buy EV/RTP reports.
15. Scope exclusions
- Pragmatic Play name, Gates of Olympus name, Zeus, art, audio, UI or copied symbol designs;
- undocumented claims about the reference game's RTP, weights or paytable;
- progressive jackpots, Hold & Win, Wild symbols or persistent cross-spin multipliers;
- changing
gates_of_olympus_stylein place; - production/stage deployment, certification claims or market enablement in the draft phase.
16. Delivery sequence
- Land reusable engine capabilities with parity and money-path tests.
- Create a new
vault-breakers-super-scoregame pack and canonical draft config. - Record config hash and backend Git identity in the game README and Math Spec.
- Run the unbalanced reference baseline and publish evidence as baseline, not target achievement.
- Tune the authored paytable and weights to 96.00%.
- Produce max-win, parity, simulation and integration evidence.
- Promote lifecycle status only through the repository's existing governance gates.