Bit Key Rush

A multi-step hacking-themed game: the player cracks a 12-character encryption key one character at a time while an FBI trace closes in.

Property	Value
Game ID	bit-key-rush
Type	multi_step
Volatility	Very High
Max Win	500× the per-guess stake, per round
Engine version	1.5.0
RTP variants	91.60% / 93.60% / 95.60% / 96.60% (best play)
RTP basis	Best play (optimal-play ceiling): no strategy returns more

Figures regenerated 2026-10-20 and 2026-10-21

From the shipped engine (v1.4.0): best play by tools/certification/best-play.mjs, the completion figure by its --completion mode, and the 10-million-round simulations by tools/certification/simulate.mjs. Engine 1.5.0 (2026-10-21) added free-chip rounds and left cash rounds unchanged; the cash figures were re-solved and are the same.

How a round works

1. The player picks a per-guess stake. Starting a round costs nothing; each guess is paid for on its own.
2. At round start the server draws a secret 12-character key: 12 independent draws from A–Z and 0–9 (36 characters), so characters can repeat.
3. The key is cracked left to right, one position at a time. Every guess costs one stake.
4. Correct guess: the character is revealed, play moves to the next position, and the guess pays at once (see Progressive payout). The FBI trace is cut by a quarter (× 0.75).
5. Wrong guess: costs an extra |wrongPenalty| × stake on top of the stake, and raises the trace by an RNG-drawn amount inside the variant's band. On touch devices the on-screen keypad dims the characters already tried at the current position, so a player never has to repeat one.
6. The round ends when:
   • the whole key is cracked,
   • the trace reaches 100 (busted), or
   • the round's credited wins reach the 500× cap.
7. The player can cash out (end the round) after any guess.

Progressive payout

A correct guess at position p (0 for the first character, 11 for the last) pays

stake × pool multiplier × payoutScale × positionRamp^p

payoutScale × positionRamp^p is rounded to six decimals, exactly as the engine computes it, and the multiplier shown on screen includes it. Later characters pay progressively more, and positions can only be cracked in order. That is what makes cracking the whole key the best play, and why cashing out early returns less.

There is no completion bonus. Cracking the twelfth character pays that guess like any other and ends the round.

Money moves per guess

• Each guess debits its stake (plus the penalty when wrong) and credits its win, if any, straight away.
• Wins are kept on a bust. A trace of 100 ends the round, but nothing already credited is taken back. What the player loses is the position they were on.
• Cash out moves no money. Wins were already paid guess by guess, so cashing out only ends the round. It never pays "multiplier × bet".
• The max win caps what one round can credit at 500× the per-guess stake (or the operator's lower maxWin). The guess that reaches the cap is paid up to it, and the round ends there.

Free-chip rounds

A round paid for with a casino free-chip offer (engine 1.5.0 and later) plays by free-chip rules, fixed when the round starts:

• Every guess costs exactly one chip, right or wrong. There is no wrong-guess penalty on a free chip.
• A correct guess pays stake × pool multiplier × chipPayoutScale × positionRamp^p: the same ramp, with the variant's free-chip payout scale in place of payoutScale.
• Everything else is unchanged: the key draw, the pool, the FBI trace, the 500× cap and the end conditions.

Aggregators book free-chip stakes in whole chips only, which a penalty of 0.47–0.61 of a stake could not be. The free-chip scale is solved exactly like the cash one, so best play on a free chip sits at the same place under the label, and every way of playing returns within a quarter of a point of the same play in cash.

Variant	chipPayoutScale	Best play, cash	Best play, free chip	Completion, free chip (exact)	Completion, free chip (10M rounds)
rtp-92	0.5295	91.599%	91.597%	90.06%	90.043%
rtp-94	0.5923	93.602%	93.592%	92.53%	92.563%
rtp-96	0.4884	95.601%	95.598%	88.66%	88.667%
rtp-97	0.5988	96.601%	96.604%	94.12%	94.130%

Round records carry roundInputs.freeChip: true; the verification page and the server-side verifier replay such rounds by the free-chip rules from the published paytable (freeChip.payoutScale).

FBI trace

Event	Trace
Wrong guess	rises by traceIncreaseMin + v/100 × (traceIncreaseMax − traceIncreaseMin), with v drawn uniformly from 0–100
Correct guess	cut by a quarter (× correctTraceReduction 0.75; engine 1.4.0 and later)
Reaches 100	round ends (busted); credited wins are kept

Variant settings

Each variant's settings live in games/bit-key-rush/src/math/rtp-XX.ts. A correct guess pays a multiple of the per-guess stake, not of a whole-round bet. The values are large because a correct guess is rare: 1 in 36 on the first attempt at a position.

Setting	RTP_92	RTP_94	RTP_96	RTP_97
Correct-guess pool (before the position factor)	6 (1) · 11 (1) · 17 (11) · 26 (32) · 39 (31) · 54 (24)	6 (1) · 11 (2) · 18 (16) · 27 (30) · 40 (33) · 54 (18)	7 (2) · 12 (1) · 20 (30) · 30 (26) · 44 (24) · 61 (17)	8 (31) · 14 (10) · 24 (10) · 37 (17) · 54 (16) · 74 (16)
payoutScale / positionRamp	0.8036 / 1.18	0.8767 / 1.18	0.7704 / 1.05	0.8681 / 1.15
Trace added per wrong guess	2.5925–4.2925 pts	2.75–4.75 pts	1.43–2.73 pts	2.5–4.2 pts
Wrong-guess penalty	0.54× stake, rounded up	0.50× stake, rounded up	0.61× stake, rounded up	0.47× stake, rounded up
Published RTP (best play)	91.60%	93.60%	95.60%	96.60%
Completion RTP (play to the end)	89.98%	92.48%	88.42%	94.00%

Pool cells list each multiplier with its weight in brackets; weights total 100. The completion RTP is the exact return of elimination guessing without ever cashing out (node tools/certification/best-play.mjs --game=bit-key-rush --completion), checked by the 10-million-round simulations in games/bit-key-rush/simulation-results/.

RTP: the published figure is best play

• Published RTP: the figure the game advertises, which calculateTheoreticalRTP returns. It is the return of the best possible strategy, solved exactly over everything the player can see (position, characters tried, trace, credited so far, with the 500× cap) by tools/certification/best-play.mjs, recorded in docs/certification/evidence/best-play.json and pinned in each variant's config. Platform policy: at or below the variant's label and no more than one point under it, so no way of playing beats the advertised RTP.
• Completion RTP: the return of elimination guessing (never repeating a character already rejected at the position) without cashing out. The engine reports it separately, and it is the strategy the simulation evidence plays.

End conditions

Condition	endReason	Money
Whole key cracked	jackpot	the last guess pays like any other; no bonus
Trace reaches 100	busted	wins already credited are kept
Credited wins reach the cap	max_win_reached	the capping guess pays up to the cap
Player cashes out	cash_out	nothing moves

Example round

The factors below are illustrative, not a real variant: stake 1.00, payoutScale 0.80, positionRamp 1.15.

Guess 1  position 0  'Q'  wrong    −1.00 stake, −0.50 penalty   trace  0.0 → 3.1
Guess 2  position 0  'K'  correct  −1.00 stake, +24.00 (pool 30× × 0.80)   trace 3.1 → 2.33
Guess 3  position 1  '7'  wrong    −1.00 stake, −0.50 penalty   trace 2.33 → 5.48
Guess 4  position 1  'A'  correct  −1.00 stake, +27.60 (pool 30× × 0.80 × 1.15)   trace 5.48 → 4.11
Cash out → no money moves; the round credited 51.60 and debited 5.00, net +46.60

TIP

A player who stops after the first character collects only the cheap early payouts. The position factor puts the value at the end of the key, but a correct guess only cuts the trace by a quarter, so playing on regardless gets traced far more often than not: best play pushes for the late characters only while the trace is low. Even best play returns no more than the variant's label.