RTP & Simulation Validation

To validate the shipped game math, a Monte Carlo harness ran the production engines verbatim (lifted from serverless/entropy.js, serverless/crash.js, serverless/lootbox.js) — 10,000 bets per Entropy entry, 0.01 ETH bets, a virtually funded 50 ETH treasury per game. Source: scripts/sim-entropy-rtp.cjs.

Results — 10,000 bets each

Game	Measured RTP	vs theory	Biggest hit	Hit paid N×	Treasury 50 ETH →	Worst dip
Mines 3	99.00%	±0.00%	1.12×	9,028×	49.00	48.91
Mines 10	98.67%	−0.33%	1.65×	6,046×	48.67	48.42
Limbo 2×	98.30%	−0.70%	2×	4,914×	48.30	47.78
Limbo 10×	101.40%	+2.40%	10×	~1,000×	51.40	49.13
Limbo 100×	89.00%	−10.00%	100x	89×	39.00	34.42
Plinko low	98.88%	−0.12%	10×	7×	48.88	48.74
Plinko med	100.19%	+1.19%	33×	5×	50.19	49.50
Plinko high	101.64%	+2.64%	170×	5×	51.64	48.21
Keno 4	95.38%	−3.62%	59.5×	17×	45.38	45.18
Keno 10	91.15%	−7.85%	237.9×	4×	41.15	38.57
Crash 2×	95.46%	−0.54%	2×	~4,770×	45.46	45.42
Crash 10×	95.70%	−0.30%	10×	957×	45.70	45.55
Lootbox	~33.5%*	by design	27.8×	6×	44.01	44.01

No treasury busted at 50 ETH with 0.01 bets. Worst drawdown was −31% (Limbo 100× — the swingiest entry). * The lootbox ETH-RTP is low by design: 94.4% of boxes pay DUST only (reactor fuel, zero treasury cost). Real treasury cost = the stock tiers ≈ 0.0006 ETH/box average → the house keeps ≈ 0.0003 of each 0.0009 ETH box. The dust economy absorbs the rest. See Lootboxes.

Aggregate house view (130k bets)

Total staked : 1,209.00 ETH      Total paid: 1,167.77 ETH
Blended RTP  : 96.59%  (excluding lootbox: 97.1%)
House kept   : 41.23 ETH         Treasury: 50 → 91.23 ETH

RTP progression insights

Low-variance games converge fast and cleanly. Mines 3 walks 97.66% → 99.00% in a near-straight line; Crash 2× locks onto ~95.5% by bet 3,000. Exactly the 99%/96% theory.
High-variance games swing hard in finite samples. Limbo 100× sat at 50% RTP for 2,000 bets (zero 100× hits — pure tail luck) before clawing to 89%. Keno 10 opened at 118.5% (an early top-tier hit) then drifted to 91%.
Tails need volume. Expect ~100k–1M bets for the extreme-multiplier entries to converge inside ±2%. This is expected statistics, not a math problem — the engines are exactly on their designed edges.
The floor2 rounding in the engines shaves a hair below theory on small-multiplier configs (visible on Mines).

What this means operationally

Question	Answer from the sim
Can a hot streak bankrupt the house?	Not at these bet caps — worst drawdown −31% with a 5,000× bankroll-to-bet ratio
Is the edge real?	Blended 3.4% hold over 130k bets vs 1–4% theoretical per game
Are payouts correct?	Harness used the shipped engines verbatim — any payout bug would reproduce here
Vault sizing rule of thumb	≥ 3,000× max bet per game keeps ruin probability negligible at these variances

Re-run it yourself:

node scripts/sim-entropy-rtp.cjs

Next: API Reference.

Overview

Payment Structures

Casino Economy

Earn & Lending

Monetization

Reference

Rtp Validation

RTP & Simulation Validation

Results — 10,000 bets each

Aggregate house view (130k bets)

RTP progression insights

What this means operationally

​RTP & Simulation Validation

​Results — 10,000 bets each

​Aggregate house view (130k bets)

​RTP progression insights

​What this means operationally

RTP & Simulation Validation

Results — 10,000 bets each

Aggregate house view (130k bets)

RTP progression insights

What this means operationally