Parcl
Parcl lets traders take long and short positions on real estate prices: New York, Miami, Dubai. The protocol has gone through four architectures. I have built through all of them, and each one taught a lesson the next one used.
v1: tokenization
We started in the tokenization era. Users deposited USDC in a vault, minted a city token, and traded it on an AMM. The market price slid far from the published index price and stayed there. Vault positions were shorts, but no user read them that way. We closed it down for a better model.
v2: isolated perpetuals
v2 added leverage on both sides, a funding rate to balance the markets, and an isolated liquidity pool per market. On paper the model was sound. We launched into the bottom of the bear market, weeks after the FTX collapse, and watched a live dashboard show zero volume. Worse, the funding rate was easy to manipulate. Programmatic traders drained buy-and-hold users, which is the opposite of what a real estate product should feel like.
v3: unified liquidity
v3 modeled its funding on Synthetix v3: a funding velocity that adjusts with skew, so staying in the crowded side gets progressively more costly. We added skew impact so single trades could not move whole markets, and we unified every market behind one liquidity pool. v3 launched in November 2023 and has done more than $5B in volume. We once celebrated a $500 trade on v2. On v3 a million-dollar day was unremarkable.
v4: our own chain
v4 is a fully onchain order book exchange with cross margin, built from scratch on a dedicated execution layer that settles on Solana. This is the version where I work across the whole stack: matching, margin and liquidation logic, validators, and indexers.
My role
Parcl hired me in 2021 to build the frontend. For v1 through v3, I built the apps and parts of the backend, and later took ownership of the v3 smart contract. For v4, I build the exchange itself.