Three entries explain the whole house.
Sowdex keeps an atelier's vocabulary because that is exactly what the protocol is: a measured basket, a tradable share, and a floor that keeps the accounts honest. No discretionary managers, no signature chains — just weights, bands and NAV.
- OneThe basket
- A surveyed set of markets — 2 to 8 legs, explicit weights, a leverage target between 1.0× and 3.0×, and the rebalance band that keeps it honest. Declared once at minting; immutable afterwards, except where governance narrows it.
- TwoThe share
- The ERC-20 that proves your share of a basket. Minting strikes shares at NAV; redeeming burns them at NAV. Transferable, listable, composable — the strategy is the token.
- ThreeThe floor
- The ground every basket stands on: an oracle registry marking each leg to market, a counterparty pool underwriting the leverage, and a backstop that absorbs bad seasons in a published order — never at discretion.
The baskets, read live from the chain.
Every basket on the floor, straight from the vault contracts — leverage against target, NAV per share, state and oracle freshness. Pick one and operate it at the Desk.
| Basket | Leverage act / tgt | NAV / share | State | Oracles | |
|---|---|---|---|---|---|
| Reading chain… | |||||
Minted at NAV · redeemed at NAV · states are deterministic health() outcomes, never an operator's call
“You hold the share.
The strategy is the token.”
From ETH to yield in four cuts.
Compose
Pick 2–8 legs from the price registry, set weights that sum to 100% (no single leg above 50%), and choose a leverage target from 1.0× to 3.0×. The composer refuses anything outside the bands — weights, leverage, leg count, all checked on-chain.
Mint
The basket compiles to a vault contract. Wrap native ETH, deposit, and the vault strikes shares at the current NAV per share. No lockup, no whitelist, no signature — two transactions on first mint: approve, then deposit.
Keep
Keepers watch the drift. When actual leverage strays 0.25× from target, anyone may rebalance for a fixed bounty. Oracles mark every leg on a 300-second heartbeat; one stale leg halts the whole floor, not just one basket — failure is loud, not quiet.
Redeem
Burn any amount of shares at NAV, any time. Exit terms are declared at minting and can only improve. Selling the share is exiting the strategy — nothing settles off the record.
The arithmetic, before you sow.
Ten wrapped ETH walk the full path — deposit, entry fee, shares struck at NAV, target exposure. Every number below is a declared parameter, not a projection; nothing here depends on an operator behaving well.
Fees accrue per second, performance by the dilution method over a high-water mark · each bound is hard-capped on-chain (≤ 50 / ≤ 20 / ≤ 2 bps×100)
The fine print, without the fine print.
| Parameter | Default | Bound |
|---|---|---|
| Weight constraint | Σ|wᵢ| = 100% | single leg ≤ 50% · long-only in Phase 1 |
| Legs per basket | 2 – 5 | hard cap 8 |
| Target leverage | 1.0× – 3.0× | creation refused above 3.0× |
| Rebalance trigger | |L_actual − L_target| ≥ 0.25× | permissionless · keeper-bountied |
| Emergency deleverage | L_actual ≥ 3.25× | anyone may call |
| Soft liquidation | NAV ≤ 15% of entry | — |
| Entry / exit fee | 10 bps / 10 bps | each ≤ 50 bps |
| Management fee | 1% / year | ≤ 2% · accrues per second |
| Performance fee | 10% over high-water mark | ≤ 20% · dilution method |
| Oracle heartbeat | 300 s (testnet) | any stale leg → protocol halt |
| Deposit asset | WNT — wrapped native ETH | 18 decimals · protocol unit of account |
| Gas token | ETH (native) | ~0.01 gwei typical |
Draft a basket yourself.
This is the exact shape the on-chain composer accepts. Step the weights in 5% increments, pick a leverage target, and read the verdict — every rule here is enforced by the factory, not by a frontend promise.
Weights shown as 5% cells · demo only — nothing is signed or sent
Who stands under the leverage?
- UnderwritingThe counterparty pool
- Liquidity providers deposit WNT and underwrite the synthetic exposure every basket runs on. Leverage never comes from borrowed assets or an off-book desk — it comes from the pool, priced by utilization, visible on this page in real time.
- Bad seasonsThe backstop
- When a basket lands badly, losses absorb in a published order — fees first, then the pool's buffer — never at an operator's discretion. The order is declared in the contracts and cannot be rearranged after the fact.
- BookkeepingAnyone, for a bounty
- Rebalances and emergency deleverages are permissionless calls with fixed bounties. There is no privileged keeper to bribe or wait for — the floor keeps itself, one bounty at a time.
Three seasons, on the record.
Testnet MVP
Six non-upgradeable contracts on Robinhood Chain 46630, two demo baskets, seven mock feeds on a 300 s heartbeat, permissionless rebalances with bounties.
Live nowDepth
Short weights, liquidation incentives, fuzz and simulation reports published with the dossier, Chainlink stock and ETF feeds mapped to the registry.
Mainnet review
Compliance assessment, audited contracts, position limits ramp, and the $SOWX utility evaluated against the checklist before anything ships.
Asked, answered.
Q·01What exactly do I hold?+
Q·02Where does the leverage come from?+
Q·03What if an oracle goes stale?+
Q·04Can the terms change after I mint?+
Q·05What does it cost?+
Q·06Is this audited? Is there a mainnet?+
The floor is set on Robinhood Chain.
- Network
- Robinhood Chain Testnet — Arbitrum Orbit L2 · chain 46630 · gas in native ETH
- Contracts
- 15 deployments — 6 core contracts, 7 mock feeds, 2 demo baskets · non-upgradeable
- Price feeds
- 7 mock feeds (BTC, ETH, NVDA, TSLA, AAPL, SPY, QQQ) · 300 s heartbeat · Chainlink path mapped
- Baskets
- MIX2X (BTC 50 / ETH 50, 2×) and TECH3X (NVDA 40 / TSLA 30 / SPY 30, 3×) live · backstop pool seeded with WNT
- Next season
- Short weights, liquidation incentives, fuzz & simulation reports
Sowdex is experimental software on a public testnet. Contracts are unaudited. Leveraged baskets can lose all deposited capital. Nothing here is investment advice.