GMX: The Trading Platform of the People, by the People, for the People
With everything moving so fast in DeFi, we sometimes lose sight of what we were trying to build, and the feelings behind why we did it in the first place. The goal should be to empower individuals first — the success of crypto shows us that tools designed by a few for themselves eventually become tools for everyone. With this in mind, I’ll be talking about a grassroots perpetual futures and spot trading platform called GMX that could have substantial value accrual in a high growth space.
Summary
The broad category of enabling the full transition of activity from centralized exchanges to DeFi is massive. We need a robust set of financial tools (derivatives, market making capacities, etc.) to become native to blockchains while remaining suitable for the same level of users and volume. GMX is one potential piece to the puzzle that enables the decentralized trading of perpetual futures on BTC, ETH, LINK, and UNI with up to 30x leverage on Arbitrum (the largest ETH L2 scaling solution) and Binance Smart Chain.
3 main factors behind the investment thesis:
- Perpetual futures are the main source of liquidity and trading activity in crypto: volume for BTC and ETH futures are over 10x greater than spot, $2T/month vs. ~$200B/month*. As regulatory concerns escalate and target CEX activities, like China forcing Huobi to stop servicing China-based users by the year end and exchanges like Binance clamping down on KYC, highly likely this volume is forced to go to decentralized derivatives. Volume is largely a function of network effects with GMX on pace to do over $2B in monthly volume which translates into roughly $4M/month in revenue from trading fees. While this is still a fraction of decentralized margin trading volume (~2%) worth mentioning that the platform has only been live on Arbitrum for 2 months and L2 solutions are one of the highest growth sectors in themselves.
- Unlike most other perpetual trading platforms, either centralized or decentralized, GMX distributes ALL of the trading fees to $GMX (the governance token) and $GLP (the liquidity token) stakers. 50% is given to GLP stakers since they take on the risk in effectively collateralizing the leverage trading system, 30% to GMX stakers, and 20% are converted to GLP and added to the $GMX price floor contract. All of these create a flywheel to increase liquidity, margin trading volume, and fees generated by the platform.
- GMX is not only a decentralized derivative platform, but an elegant design to gain liquidity to swap multiple DeFi assets with no slippage (up to a certain size, at least) using GLP pools. While unclear if an outright better model at the moment, GLP is essentially a shared pool liquidity mechanism (like SNX) used to collateralize all traders on the platform that allows traders to happen with no slippage at the direct oracle price. Inclusion of the index model, which now over $70M in liquidity, into DEX routers could drive spot volume to a currently underappreciated aspect of the platform and make it competitive to trade certain sizes of large assets vs. AMM’s like UNI, SUSHI, CRV, etc.
The Market Opportunity
As a quick refresher, perpetual futures (will refer to these as “perps” from now on) were pioneered by the BitMEX team a number of years ago but really exploded in popularity last summer. They are effectively futures contracts that never expire but market participants pay/receive some sort of fee, called a funding rate, to keep on. Negative funding rates means you have to pay to be short (you get paid to be long) and positive funding rates mean you pay to be long (get paid to be short). If the index price of the perp is higher than what the spot price of the asset is, funding will be positive (i.e., to arbitrage back to the “real” price, we need to incentivize people to short and penalize longs) and if the perp is trading lower than spot, funding will be negative (penalize shorts and incentivize longs).
While perps on centralized exchanges do $2T+/month regularly, decentralized derivative platforms only do ~$100B/month.
Even if perps as an instruments don’t attract more interest which is highly doubtful given the institutionalization of the asset class, there are multiple reasons why more of this centralized volume could move to decentralized platforms. They include, but are not limited to:
- Regulation limits leverage and margin taking ability on centralized exchanges or makes on-ramps to those products too difficult
- Market participants doing similar activities on DeFi protocols are rewarded more for the same activities, most popularly through liquidity mining programs (see DYDX) and trading fees are typically distributed in some form to stakers of the native token instead of going to the exchange (except for DYDX, whose token is controversially only governance and trading fee reduction, equity holders get a share of rading fees instead)
- A large reason decentralized derivs haven’t taken off yet is due to the difficulty of blockchain scaling but that is longer an obstacle
- New innovative features (like GMX’s GLP or Perp’s Private Markets) create flywheel effects that drive adoption
GMX Overview
GMX is a spot and derivatives platform with market and limit order functionality to trade various large-cap assets with up to 30x leverage and no slippage. This is done through a shared liquidity mechanism called GLP, a pool of all tradable assets on the platform. Since there is no orderbook, trades can be made at the current oracle price (secured but also limited by Chainlink price feeds) with theoretically infinite depth for all supported assets. There are two parts to the GMX system: the GLP token for trading liquidity, and the GMX token for governance.
GMX started as Gambit Financial, which was only on BSC, but rebranded to GMX earlier this year with the expansion to Arbitrum which now represents almost all of the margin trading volume. Nearly all the spot volume on the platform still occurs on GMX BSC though, given native integrations with DODO for smart order routing through GLP.
As of writing, the platform has done roughly $2.8B in total volume and collected ~$3.9M in fees since September 2021. Fees are 10 bps to open/close a trade and a borrow fee (a funding rate) based on open interest/the utilization of assets in the GLP — more on this later.
Annualized using the most recent week as a run rate, the platform generates ~$50M in fees/year, with half of that paid to GLP holders in ETH (implying GLP trades 3x P/E, 33% apy), 30% to GMX stakers (8x P/E, 6% apy), and 20% directly into GLP for a GMX price floor fund.
Out of the two components here in GLP and GMX will talk about GLP first since that’s really what makes the engine run.
GLP: GMX’s Liquidity Provider token
GLP is an index of the large-cap assets (think SP500) supported by the GMX protocol that auto-rebalances every week. This is used for swaps and for traders to margin trade — when leverage traders win, GLP holders lose, and when traders lose, GLP holders profit. As of writing, it consists of ETH, BTC, LINK, UNI, USDC, and USDT, and has roughly $73M in AUM (has doubled over the past month).
You can buy/mint GLP out of any of the supported assets, but fees will be lower for tokens that the GLP has less of compared to its target weight. For example, looking at the image above, ETH, USDC, and USDT are above its target weight in the GLP already (so there will be higher fees) while the other assets are below, and will have no fees to mint GLP from those assets to incentivize users to rebalance the pool themselves. We can confirm these fees on a 100 ETH mint of GLP below:
One last thing to discuss here is utilization. Utilization represents the amount of the asset that is being used to take a margin position at the moment on GMX; at 100%, that asset can no longer be used to open any more positions. You can think of this like a cap on a funding mechanism; ETH utilization is so high right now for example (87%) because people want to open their ETH longs with ETH. There is currently a proposal underway (see here) to increase liquidity to expand trading capacity. Stablecoin utilization rates have been close to 0% since October but were higher in September, corresponding to what we’d see as stable-margined funding rates on centralized exchanges.
Risks of GLP are not insignificant; in the case that traders win massively, holding GLP could hurt you — the biggest reason why most of the fees are shared with these holders over GMX stakers. However, if we look at the data (and anecdotal evidence from leveraged traders in general) traders normally have negative pnl outside of raging bull markets. On GMX through the first week of October traders were in the red but as BTC and ETH rallied 20–30%, traders on the platform have made substantial amounts of money (nearly $3M).
GMX: GMX’s Utility and Governance Token
GMX is used in governance and can be staked to receive 30% of the total fees generated by the platform, delivered in ETH. An additional 20% of fees is converted to GLP and deposited into the floor price fund, which is roughly $6M right now. While insignificant at the moment, as fees increase this creates a virtuous cycle further bolstering the system.
Rewards are also given in escrowed GMX tokens that vest to become actual GMX tokens over the next year. Long-term stakers are incentivized to remain staked through multiplier points, a gamified system that rewards you points for staying staked for longer and burning when you unstake. More on the rewards system here: https://gmxio.gitbook.io/gmx/rewards
Tokenecon & Valuation
The max supply of GMX is 13.25 million but more could be minted through governance if required for new projects. While less than 7 million of the supply is circulating, for valuation projections we will conservatively assume the entire supply is unlocked given that we don’t know exactly when this supply will come on the market. The vast majority of the unlocked GMX is reserved for vesting from escrowed GMX rewards (which boost base APY).
Annualizing the most recent week of revenue has GMX trading at 16x fully diluted (~$50M/year), a 20% discount to it’s main competitor DYDX which trades 20x. However, would note that the DYDX token doesn’t actually give holders any share of protocol revenues, which both GMX and GLP do.
GLP trades at 3x P/E, which would make it the most undervalued token by earnings among all the tokens tracked by Token Terminal, but maybe more understandable given price appreciation is constrained by the performance of the large cap index vs. other more speculative assets.
A simple valuation framework starting with the total perpetual futures volume in crypto and GMX share within the eventual DeFi volume supports a price target of $130 at end of 2022 using a similar multiple to DYDX. This only factors in trading fees from margin trading; adoption of GLP and swap volume through router integration would be potential upside. Reasonable to expect GMX to trade at a higher multiple than DYDX to support lopsided growth prospects if the GLP model proves to be sustainable.
Competitors
90%+ of decentralized margin trading volume is from DYDX with PERP and GMX battling for the #2 spot over the past week.
GMX is the only large actively trading perpetual protocol that employs a shared liquidity model which allows trades to be executed at the oracle price with no slippage. Compared to other mechanisms, this has clear advantages including:
- Token flywheels are the most significant of all models driven by a a highly scalable GLP liquidity model: the flywheel of more GLP liquidity means larger possible trade volumes, which results in more fees for GLP and GMX holders, and creates more incentives to continuously mint GLP
- Simple to list new tokens, limiting factor is just Chainlink oracle support
However, the introduction of GLP as liquidity also has some drawbacks, mostly around long-term viability including:
- Unclear if a shared pool model is strictly better than an orderbook or AMM model
- Liquidity providers are at risk of traders going on a roll and bankrupting the system, or an oracle exploit
- ETH APY for GLP holders is >25%, which should be enough to compensate for these risks, but 1-yr vested tokens represent 70% of the advertised APY
- To avoid bots from taking advantage of the lag in oracle pricing vs. centralized exchanges, if you want to close positions in <24 hours, you need to have a price change of at least 1.5% (this doesn’t exist on other platforms, and will limit volume from hft arb/basis strategies)
Conclusion
There are many structural headwinds supporting an increase of trading activity on DeFi that GMX is well-positioned to take advantage of in the near future. Unlike many other applications of DeFi that are speculative in nature, perpetual futures have repeatedly shown product market fit, running over $2T/month in volume on centralized exchanges.
The lack of effective infrastructure was the biggest reason in my mind for not seeing decentralized derivs adoption but now that time has passed with L2’s coming into the spotlight. Nearly all of these platforms were unusable on L1 blockchains due to high transaction costs and a lack of infrastructure to fully support them. Now that we have this in place, these are generating significant amounts of revenue that can no longer be ignored compared to their centralized counterparts.
Important to realize that the pie here is so big that despite DYDX’s well established advantage and competition from new AMM models, all GMX really has to do is capture <5% of DeFi margin trading volume to continue generating substantial value to token holders. Excited to see the GMX flywheels in full action securing the long-term viability of GLP and more attention on the protocol with the recently launched trading competition.
BlockTower is long GMX and some of the related assets mentioned in passing in the article. Full disclaimers here.
