Skip to main content
No. Fade is a payout layer: a shared pool, a randomness source and one arithmetic check. It does not offer anything to the public itself. Apps build products on it: checkout win-backs, swap promotions, rewards with upside, and games. Games are one use case among several.
A VRF gives you a fair draw. It does not give you the money to pay a fixed-odds promise. “Every order has a 1 in 100 chance to be free” must pay even if three people win in a row on a quiet day. Fade adds the capital that carries that variance, shared across every app, with on-chain rules for how much it can risk.
No. An app is whoever signs open_wager. Its record is created by its first wager. There is no deposit, no bond and no allowlist. What bounds an app is the pool’s caps: its open liability can never exceed 20 % of the pool’s assets.
The full paytable is an argument of open_wager and is copied into the wager account before the randomness exists. Anyone can read it in an explorer and check that the app’s interface matches it. The program refuses any paytable whose expected return exceeds 1 − edge, in exact integers. See Randomness for how to verify a draw.
No. A wager whose worst case does not fit the pool’s caps is refused at open. A wager that was accepted has its worst case reserved, and settlement pays it in full. If the beneficiary’s account cannot receive, the payout is held in escrow and delivered later.
After T_SETTLE (150 slots, about a minute), anyone can expire the wager. The pool keeps the stake. That rule is deliberate: returning stakes on a missing draw would let anyone who can withhold draws cancel their losing wagers for free. The trade-off is that a long oracle outage costs the wagers in flight.
Anyone. Both are permissionless and prepaid by the wager: 50 000 lamports to request, 200 000 to settle. Fade runs a public keeper at api.fade.finance, and anyone can run their own. See Settlement and keepers.
Because nothing is refunded. In a checkout win-back the sale stands, the buyer keeps the item, and the pool pays out an amount equal to the purchase. The buyer won it back.
The pool’s share of the edge on every stake: at least the 100 bps LP floor, plus whatever edge the games leave above the fees. The result depends on volume and on luck, and months can end down. Fade publishes no headline yield. Read Risks.
A withdrawal request clears at the next strike, after the epoch closes and every wager opened before the close has resolved. Capital reserved by open wagers cannot leave, so under heavy use a withdrawal can be paid over several epochs. Shares are transferable SPL tokens and can also be sold to anyone willing to buy them.
On the direct path, the protocol’s reference minimum is 7.5 USDC: below it, the fixed SOL cost of a wager’s own oracle draw would exceed what the house expects to earn. Devnet runs with 0.10 USDC. Stakes of a few cents are what the shared beacon design is for.
Legacy SPL USDC. Token-2022 is rejected, because a transfer hook would give a third party a call into every payout.
No instruction can rewrite a balance, override an outcome or move pool funds outside the user paths. The program is upgradeable, which is a real power; on mainnet it sits with a multisig council behind a 72-hour timelock, with new wagers paused before the clock starts. A guardian can only tighten. See Security model.
app.fade.finance, on Solana devnet with test USDC from the built-in faucet. See Try it.