What are we building

GTE is the next-generation spot exchange: a lightning-fast, fully on-chain central limit order book (CLOB) + automated market maker (AMM) on MegaETH offering the best trading experience across all of crypto.

Liquidity is a necessity for blockchains to allow users to frictionlessly exchange assets and bring new financial primitives on-chain. AMMs bootstrap liquidity well for exotic assets, but the incentive structure for passive liquidity is not sustainable for most crypto markets.

MegaETH will be the world’s fastest real-time blockchain. We are leveraging its performance to build the first fully crank-less, on-chain order book for active liquidity within the Ethereum ecosystem.

Our team comes from backgrounds in HFT, crypto infra, and fintech. We are building GTE in close collaboration with the MegaETH team through their flagship incubator program.

CLOB

Markets exist when there is sufficient demand to buy/sell an asset along with a counterparty willing to facilitate its exchange—the market maker. Traders rely on market makers to provide liquidity, who expect to profit from their service in return. Since asset prices constantly change, market makers need an exchange structure that lets them frequently update their posted liquidity based on their view of prices. Central limit order books (CLOBs) are the natural place for market makers to provide active liquidity. ~97% of daily crypto spot trading happens on CEXs (CLOBs) like Binance vs. ~3% on DEXs (AMMs) like Uniswap, especially among blue-chip assets. Traders often prefer centralized exchanges since deeper liquidity from market makers results in less slippage, especially for larger trades.

Limitations of passive liquidity

Market makers currently opt for trading on CEXs because DeFi AMMs are hindered by passive liquidity provision. While AMMs provide continuous liquidity on-chain by market-making automatically, their design makes LPs susceptible to well-researched forms of loss, mainly loss-versus-holding (impermanent loss) and loss-versus-rebalancing (LVR). When the price of an asset increases, the ratio of the assets inside the AMM pool changes accordingly (based on the bonding curve, e.g., xy = k). While the net value of the LP position also increases, the LP ends up worse off than had they simply held without depositing into the AMM. If the prices end up reverting back, the LP recaptures this loss (hence impermanent), yet in any scenario, the LP suffers from LVR. LVR occurs when price discovery happens on faster CEXs like Binance, so LPs are prone to adverse selection from stale prices in the AMM—they are on the wrong side of trades that capture the CEX/AMM arbitrage. AMMs always trade at prices worse than CEXs (for LPs) because they do not reactively update based on price changes on other exchanges. A rebalancing portfolio (replicating the same LP trades but on a liquid CEX) would otherwise capture these profits, which are quantified as LVR. Even on concentrated liquidity AMMs where LPs can specify price ranges for their liquidity (approximating limit orders), managing positions dynamically between ranges is expensive and challenging to automate precisely. Swap fees don’t sufficiently cover these losses for LPs, so market makers will always choose to post active liquidity on order books instead of deploying passive liquidity into AMMs.

Requirements for active liquidity

A significant bottleneck in bringing order books with active liquidity on-chain is gas. Market makers profit from the bid-ask spread on each trade when trading on order books, which they maximize by trading at high frequency. Since prices are constantly changing, they need to frequently cancel/update orders to avoid adverse selection—they will get picked off (filled on a quote they are trying to cancel) if they don’t cancel stale quotes quickly enough. On traditional finance exchanges, canceling orders is a free operation. On a fully on-chain order book, however, canceling an order requires gas due to state changes, and gas is too expensive, especially within today’s Ethereum ecosystem.

Current order book designs often feature a crank separating order matching and settlement into two different transactions to reduce the burden of high gas costs. When a trade occurs, the order book state is updated, and the matched order gets added to a queue in a pending state. A cranker (off-chain server) must execute a separate transaction to actually settle the transfer of tokens resulting from the trade. This way, a taker filling multiple maker orders doesn’t have to pay extra gas. Adding a crank, however, creates an unnatural trading experience, breaks composability, and adds trust assumptions. Market makers have to account for unsettled account balances within their strategies, and traders must rely on a cranker to receive their tokens. Integrations with other DeFi applications become complicated since the transactions aren’t atomic within the same block. In traditional finance, multiple fills are created if a market participant trades through multiple levels. Gas should be cheap enough on-chain such that a taker can replicate the same behavior without using a crank (since gas is cheap enough for multiple fills at once).

On-chain order books also require low latency. If block times are long, prices are more likely to diverge from true prices because CEX exchanges are much faster, and most price discovery for blue chip assets happens there. Since sophisticated actors will know the true price, market makers become more susceptible to getting picked off because they cannot reactively adjust their quotes quickly enough. A low latency chain reduces the likelihood of market makers finding themselves in such scenarios.

Ultimately, marker makers maximize revenue, so a fully on-chain order book only works in practice if the underlying chain has cheap gas and low latency. The expected profit they capture from spreads (good trades) must sufficiently outweigh the operational costs (gas fees) and the risk of getting picked off (bad trades). Higher expected profits lead to highly liquid markets with tighter spreads, which create a superior trading experience for users with lower slippage.

Why we’re building on MegaETH

MegaETH is the first blockchain to support a fully on-chain order book with performance comparable to centralized exchanges at 100k TPS and 1ms block times. Extremely cheap gas enables a crank-less design and liquid markets where makers aren’t constrained by the costs of placing/canceling orders, and settlement is instant. High throughput and low latency support many trades per second, reduce the likelihood of adverse selection, and create price discovery on-chain, averting significant LVR on both the AMM and CLOB. Since the sequencer is centralized with FIFO ordering, the matching engine can run in true price-time-priority, the same way it’s done off-chain. Since MegaETH is an L2 rollup, the order book’s EVM compatibility also makes it maximally composable within the Ethereum ecosystem. Leveraging these unlocks, GTE will create the best trading experience across all of crypto for market makers and traders, spearheading a new wave of liquidity on-chain.

Matching engine mechanism

GTE runs a fully-on chain matching engine, taking advantage of MegaETH’s FIFO, centralized sequencer for true price-time priority. The order book essentially works the same exact way as all centralized exchanges where the lowest ask and highest bid are filled first. Consequently, market makers are incentivized to compete on price and quote tighter, resulting in better execution and lower slippage for traders.

At the most basic level, all markets define base and quote assets, lot size, and tick size. Orders are added at the chosen price level (a multiple of tick size) with a specified quantity (a multiple of lot size).