Methodology

HouseMath does the house's math. Every number on this site is computed from public on-chain state — never fabricated — and labelled with the block or snapshot it came from. Here is exactly how we get each figure.

Core metric

Expected value per pull (EV)

For a gamified draw, the expected value of a single pull is the probability-weighted average of every possible outcome's value, minus what the pull costs. We enumerate the reward distribution from the contract's on-chain configuration and current backing, weight each outcome by its draw probability, and subtract the pull price:

EV_per_pull = Σ ( p_i · value_i )  −  pull_price

Where draws are backing-weighted, each outcome's value is derived from a harmonic-mean weighted measure of the backing that supports it — this properly penalises thin, easily-drained reward tiers rather than letting a single large tier dominate a naive average. A negative EV means the pull is expected to lose value at current state; positive means the reverse.

Derived

House edge

The house edgeis the share of a pull's value the contract structure retains on average — the negative of EV expressed as a fraction of the pull price:

house_edge = − EV_per_pull / pull_price

A 12% house edge means that, on average, a player gives up 12% of what they stake per pull. This is the analogue of a casino's built-in margin, read directly from the chain.

Structural

Health metrics

Exit-liquidity risk measures how exposed players are if the backing liquidity behind rewards were withdrawn — a normalised 0–100% figure where higher is riskier. House-edge pressure restates the edge as a health gauge. These are structural reads, not predictions of any individual outcome.

Risk audit

Grades & findings

The per-protocol risk grade (A–F) and the published audit combine the economic metrics above with a manual structural review of the contract — admin powers, upgradeability, liquidity mechanics, and reward accounting. Findings are labelled by severity, category, and status. This is independent analysis, not a formal security audit.

Provenance

Data freshness

Live headlines are read on demand and cached briefly, then labelled with the block number and time they came from. When a live read is momentarily unavailable, we fall back to the most recent stored snapshot and label it as such — and if neither is available, we show “headline updating” rather than a stale or invented number.