I've been asked a few times about how our RFQ systems work. Perps, spot, and soon-to-be HIP4 markets use different mechanisms. Spot is a simple OTC-style transaction, but the perps side is more interesting. We want to give users a firm price in an order book that prioritises cancellations. This leaves some traders unsure how much liquidity is actually there when they place a large order, so traders instead use TWAPs or drip their orders using another method. This theme recurs across most of our conversations. We think a better approach is to let makers and takers pre-negotiate a price for a given size. The next problem is how to open or close at that price in an order book. Some solutions temporarily flood the books or have users DM MMs to increase liquidity until their order executes. While this works in practice, it gives the user less guarantee of an end fill and exposes that liquidity to other traders. Plus, it's a coordination pain. Our solution is based on two principles: i) give the taker the best price and ii) give makers the ability to respond to large orders using the same margin they currently have on Hyperliquid For the taker's best price. If the best price is inside the spread or at the tip of the book, then the trade executes as expected. If the trade is outside of the spread but the given size would walk past that price, the maker's order acts as a price wall for the taker. So the taker gets the benefit of any liquidity between their agreed price and the book's tip. For the maker, the percentage of their order that fills depends on how close they are to the book, the order's relative size, and available liquidity. The basic heuristic is that the closer you quote to the book, the more of the order will fill. We designed the system so makers can use the same wallets they typically use to trade. One interesting use case is using the RFQ to unwind positions more effectively. The system is designed to be more deterministic for both makers and takers, providing clarity for trades that would move the book. The order book can typically absorb small/medium punts on liquid pairs. But as trade size increases or markets get less liquid, our mechanism becomes far more effective. We're working on extending the request model to support multi-leg, multi-instrument requests. This would give a better user experience for the crazier strategies I've seen out there, and with Hyperliquid's depth, it is the best place to support this use case. In the meantime, you can expect more pairs and bigger clip sizes shortly. Go trade: app.silhouette.exchange/rfq