GMX is a decentralized perpetuals and spot exchange on Arbitrum and Avalanche that pioneered the “real yield” narrative in DeFi — offering genuine ETH/AVAX token rewards from actual trading fee revenue rather than inflationary token emissions. GMX’s original model (v1) used a single shared liquidity pool called GLP — a basket of assets (ETH, BTC, LINK, UNI, stablecoins on Arbitrum; AVAX, BTC, ETH, stablecoins on Avalanche) — where depositors provided the counterparty liquidity for all GMX traders simultaneously. GLP holders earn 70% of all GMX protocol fees (trading fees 0.05-0.1%, borrow/funding fees from open positions, and liquidation fees) paid in ETH (on Arbitrum) or AVAX — genuine native token yield, not GMX inflation. GMX stakers (staked GMX) earn 30% of all fees plus esGMX (escrowed GMX with 1-year vesting, reduceing sell pressure). This model made GLP the “house” in a casino sense: traders collectively lose more than they win (expected outcome), and the house (GLP) collects both the losers’ positions AND the trading fees. Through the 2022 bear market, GLP delivered 20-40% APY in ETH at a time when most DeFi protocols’ yields collapsed — making GMX the defining real yield protocol. GMX v2 (2023) introduced isolated GM pools (individual market pools per trading pair) with Chainlink’s low-latency oracles and more sophisticated funding rate mechanics to replace the monolithic GLP model.
Key Facts
- Protocol: GMX
- Networks: Arbitrum (primary), Avalanche
- Categories: Perpetuals DEX, Spot DEX, Real Yield
- Governance token: GMX
- Liquidity token (v1): GLP (multi-asset pool)
- Liquidity token (v2): GM (per-market pools)
- Fee split: 70% to GLP/GM holders, 30% to GMX stakers
- Yield type: Real yield — ETH/AVAX, not GMX inflation
- Peak TVL: $700M+ (2022-2023)
The GLP Model: House vs. Trader
GLP = protocol-owned counterparty liquidity:
- Trader longs ETH with 5x leverage → GMX is short ETH (via GLP)
- If trader profits → GLP pays the profit
- If trader loses → GLP receives the collateral
- Fees → always to GLP regardless of trader P&L
House edge economics:
- Trading fee: 0.05-0.1% on open + close
- Borrowing fee: hourly fee for holding open positions
- Liquidation fee: position collateral → GLP on liquidation
- Net result: GLP expects positive return historically (traders lose more than they win)
GMX v2: GM Isolated Pools
GMX v2 replaced the monolithic GLP with per-market GM pools:
| Feature | GLP (v1) | GM Pool (v2) |
|---|---|---|
| Assets | Multi-asset basket | Single market (e.g., ETH/USDC) |
| Risk | All assets correlated | Isolated market risk |
| Oracle | Chainlink standard | Chainlink Low Latency |
| Funding | Fixed borrow rate | Dynamic funding rates (OI-balanced) |
| Capital eff. | Mixed | Higher per market |
Real Yield Performance
GMX historical fees (Arbitrum):
- 2022 (bear market): $80-100M total fees collected
- GLP APY: 20-40% in ETH during bear market
- GMX staker APY: 5-15% in ETH
Why it mattered: During 2022 collapse, GMX GLP generated ETH yield while:
- Most protocols’ APY: collapsed to 0-1% (emissions dried up)
- CeFi yield (Celsius, BlockFi): collapsed completely
- GMX: consistent 30%+ GLP APY from real trading activity
Related Terms
Sources
- “GMX: The Protocol That Defined ‘Real Yield’ in a Bear Market” — Messari / GMX Deep Dive (2022-2023). Comprehensive analysis of GMX’s emergence as the defining real yield protocol of the 2022 bear market — examining the GLP mechanics (multi-asset pool as counterparty), why the model generates genuine ETH yield independent of GMX token price, the marketing of 30-40% GLP APY as ETH (not GMX) in a zero-yield environment, and how GMX achieved $500M+ TVL while the broader DeFi market was in collapse.
- “GLP Mechanics: Risk, Return, and the Counterparty Liquidity Model” — Delphi Digital / GLP Risk Analysis (2022). Technical analysis of GLP liability structure — examining how GLP’s multi-asset composition creates both long and short exposure to its constituent assets (GLP is net long crypto when crypto price rises, losing if traders are also long; GLP benefits when traders lose), the delta exposure of GLP positions, the specific risk of a large directional trader winning against GLP, and how GMX manages this risk through position limits and market-wide open interest caps.
- “GMX v2: Isolated Markets, Funding Rates, and the GM Pool Architecture” — Blockworks / GMX v2 Research (2023). Analysis of GMX v2’s major redesign — shifting from the monolithic GLP to isolated per-market GM pools, adding dynamic funding rates (hourly rate shifts when long/short OI is imbalanced, incentivizing OI equilibrium), leveraging Chainlink’s Low Latency Oracle for faster price updates to reduce oracle manipulation risk, and how these changes addressed GLP’s limitations (correlated market risk, oracle lag).
- “GMX Competitors: The Perp DEX Arms Race on Arbitrum” — Token Terminal / Arbitrum Perp DEX Research (2024). Analysis of the competitive landscape for decentralized perpetuals exchanges on Arbitrum — examining GMX’s competition from Gains Network (gTrade, synthetic leveraged trading with gDAI vault), Vertex Protocol (order-book hybrid), Hyperliquid (on-chain order book), Jupiter Perps (Solana), and whether GMX’s GLP/GM model maintains competitive moat as the perp DEX landscape matures.
- “GMX Tokenomics: GMX, esGMX, and the Long-Term Staking Design” — DeFi Research / GMX Token Analysis (2023). Analysis of GMX’s tokenomics — the GMX governance token and its staking model (stake GMX → earn 30% of protocol fees in ETH/AVAX + esGMX), esGMX’s role in reducing sell pressure (1-year vesting converts esGMX to GMX), the multiplier points system (bonus points for long-term staking = compounding rewards), and whether GMX’s emissions model creates sustainable long-term tokenomics as emissions decrease.