Simulator
The physics of a 1.5% edge.
A Monte Carlo on the LP-side return. You set the assumptions about who shows up, how often they bet, what odds they pick, and how the platform grows. The simulator runs the same payout math the contract enforces on-chain, draws bet outcomes from the bettor's stated probability, and reports the median APY, the 5/95 percentile envelope, and the worst peak-to-trough drawdown the LP endures along the way. Every assumption is a slider. Reproducible from the seed.
Method
What the simulator is and isn't.
The payout math is exact. Every simulated bet computes grossPayout = stake × (1 − edge) / winProb — the same line of Solidity that runs on-chain. The per-bet LP drawdown cap (1% of net assets by default) is enforced the same way: if a bettor's stake would breach the cap, the simulator clips it instead of reverting.
The outcomes are fair. A bet at probability p wins with probability p. The simulator does not model bettor skill, alpha, or adverse selection — bettors get exactly the variance they pay for. If you believe a subset of bettors is systematically better than random at picking p, the LP yield will be lower than the simulator says; if worse, higher.
The behavior is assumed. The hard part is not the math but who shows up. The sliders encode the assumptions you have to make about bettor flow, stake size, and odds preference. Move them and watch what changes; that's the point.
The convexity is a model. The viral coefficient captures the second-order dynamic that big winning events bring more LPs and more bettors, which makes more big events possible. It is a feedback loop, not a forecast. Setting it to zero gives you a flat regime; cranking it up gives you the optimistic regime. Reality is somewhere in between, and the answer depends on whether the product crosses the network-effect threshold.
The seed makes it reproducible. Two people with the same parameters get the same chart.