Lately, the DeFi space has emphasized the importance of long-term viability and profitability. This may be a shocking concept to some, but it's natural for protocols to be viewed in a similar way to traditional businesses. Thankfully, the old model of extracting value through inflation is fading away, and projects are now focused on providing actual value to their users.

GMX is a great example of a protocol that has embraced this philosophy. It has been available for just over a year but has already established itself as a major player in the space thanks to its impressive fee generation and TVL. As with many great projects, its success is the result of a combination of elegant design and innovative thinking. But what sets GMX apart? Let's take a closer look.

What is GMX and how does it make money?

GMX is a perpetual exchange that runs on the Arbitrum and Avalanche blockchains, with the majority of activity happening on the former. Perpetual exchanges are platforms for trading perpetual futures contracts, which are like regular futures but don't have an expiration date. With GMX, you can speculate on the price of an asset without actually owning it.

GMX has two tokens: GMX for governance and GLP for pooling assets from lenders. The GLP token is an index of blue-chip tokens that consists of about half stable assets and half risky assets. Users can deposit supported tokens into the pool in exchange for GLP tokens. The current index composition and target weights on the Arbitrum chain (which will be used for all future references) are as follows:

A crucial feature around which GMX is built is that, unlike traditional DEXes that rely on the constant product market maker model (x*y=k), GMX utilizes Chainlink oracles to feed price data from centralized exchanges, enabling trades with no slippage. While this is a positive for traders, it also leaves the protocol vulnerable to price manipulation and limits its scalability potential. We will delve into these issues later on.

GMX incentivizes participants to deposit underweight assets or withdraw overweight assets by structuring the fees accordingly - it’s less expensive to deposit underweight assets or withdraw overweight ones than another way around. This is the first way the protocol generates revenue - through the minting and burning of GLP tokens.

The protocol can also function as a DEX, where GLP index tokens are traded. To incentivize users to maintain the optimal asset mix in the pool, GMX adjusts trading fees based on the desired balance of assets. This is the second way the protocol makes money.

The third way GMX generates revenue is through margin trading. The protocol charges a fee of 0.01% on the size of the position whenever a position is opened or closed. Additionally, GMX has a borrow fee of 0.01% per hour, which is calculated based on the ratio of borrowed assets to total assets in GLP. When a position is closed at a loss or liquidated, the resulting profit goes to the protocol. This is where the majority of GMX’s earnings come from, so let’s take a look at those in more detail:

Since its debut in September 2021, GMX has generated an impressive $40 million in profits (on Arbitrum alone and not including fees). When you factor in the fees, the protocol has been wildly successful, with a total haul of $135 million on Arbitrum and an additional $31 million on Avalanche.

The generated earnings are then shared with GLP and GMX token holders. The protocol rewards GLP holders with a 70% cut of the fees they earn, which can translate to APR in the 20-40% range (depending on the amount of fees collected in a given week). It is not discouraging that the rewards are paid in wETH/AVAX and escrowed GMX. GMX token holders receive the remaining 30% of the accrued fees, which amounts to approximately 10-20% APR.

GMX is self-funded, so no VCs are lurking around ready to dump on retail. The protocol is designed to prioritize the community, with the distribution that allocates only 2% of the GMX tokens to the team (250,000 out of a total of 13 million). As for the future, the number of tokens in circulation may increase beyond the current 13 million if the DAO votes in favor of it.

As a GMX token holder, staking your tokens not only allows you to earn rewards but also grants you the opportunity to participate in governance and rack up Multiplier Points to enhance your rewards (similar to veCRV in Curve Finance). While the esGMX tokens are locked up for a year, they can still be staked to earn rewards just like regular GMX tokens.

Holding GMX tokens is essentially a bet on the continued growth and adoption of the protocol, without exposing investors to the risks of being a counterparty to traders via holding GLP. However, if something were to happen to GLP, the value of GMX tokens would plummet in no time.

Risks for GMX and GLP holders

While the protocol is well-thought-out in terms of earnings and rewards, are there any design flaws that could hinder its scalability or cause it to crash? To answer this question, we need to dive a little deeper into how GMX operates.

As previously mentioned, the protocol relies on Chainlink oracles to feed asset prices to its ecosystem. This eliminates slippage but also exposes GLP holders to market manipulation. In early 2022, a trader on Avalanche exploited this vulnerability by opening and closing a series of long/short AVAX positions on GMX while simultaneously manipulating the price of the coin on centralized exchanges. This scheme resulted in a loss of approximately $500,000 for GLP holders, although it's possible that the attacker didn't profit at all due to the overall costs incurred. The GMX team responded swiftly by imposing limits on AVAX's open interests, though this isn't exactly the embodiment of a trustless protocol.

Might something akin to the AVAX incident occur on Arbitrum? It's not out of the realm of possibility, although the greater liquidity of ETH and its larger market cap make it less prone to price manipulation.

Another potential danger for the GLP pool is during times of extreme volatility. This could occur in the following way:

1. Multiple leveraged short positions are opened during a market downturn (resulting in a large net open short position in the protocol).

2. A significant piece of negative news hits the market.

3. The swift decrease in the value of risk-on assets in the liquidity pool causes the price of GLP to plummet.

4. The leveraged shorts’ profits are paid out from the stablecoins in the pool.

GLP is designed to fully back any open positions in GMX. As such, it's not possible to open a short position with more capital than is available in the pool (so that when all positions close, the protocol will still be able to pay out the rewards to all successful traders). This feature, combined with the restriction on short open interest, should prevent such an attack from occurring or having an outsized impact.

However, imposing a cap on open interest may present scalability issues. If the platform's GLP deposits increase while GMX is simultaneously capping short interest, the fees generated from trading will be limited, resulting in reduced returns for GLP holders and leading to the withdrawal of the funds from the pool.

A partial solution to this problem may be the introduction of something akin to a funding rate used on centralized perp exchanges. The funding rate is a fee paid by one side of a trade to another to ensure that the price of the contract stays aligned with the underlying asset. In the GMX context, if there is an excess of short or long open interest, the funding rate can be adjusted to even out the net difference in the positions.

One way to monitor the risk of the GLP pool is to keep track of net open interest on the platform. If the stablecoin portion of the pool is significantly greater, in USD terms, than the net short open interest in the protocol, GLP can be considered healthy. For example, if traders are net short USD 25 million and the stablecoins deposited in GLP are worth USD 200 million, the ratio would be 1/8, which would typically be considered healthy. The threshold for what is considered unhealthy may vary based on an individual's risk tolerance; for my part, I would consider leaving GLP if the ratio reached around 1/4. It's worth noting that liquidity providers can withdraw stablecoins up to the amount of the open short interest.

Since it is vital to keep an eye on the health of GLP, feel free to use my Dune query to check on the health of GLP here.

The query shows the current short, long, and net open interests along with the amount of stablecoins in GLP. The Health Factor is the ratio of net open interest (calculated only if negative), to the USD value of the stablecoins in the pool. Using the numbers from my above example, the factor of 12,5 should be safe, but if it goes up to 25, I would be thinking about getting out. This is a hypothetical scenario and most likely, if short interest were to run amok, the team would step in to tame it.

Glance at valuations

This article begins by highlighting the shift in the crypto industry towards a focus on financially sound, long-lasting protocols. Let’s then see how GMX performs when evaluated using some crucial metrics. For the sake of comparison, we will also assess dYdX, a competing crypto perp exchange that, much like centralized exchanges, follows the order book model. This model involves matching buy and sell orders and using a funding rate to balance the open interest between long and short contracts.

In December 2022, both protocols had a similar TVL of $460 million. However, over the past six months, GMX has generated $59 million in earnings, while dYdX has made approximately half of this at $38 million. That is what you call capital efficiency.

When looking at the price-to-sales ratio on the fully diluted market cap over the same time frame, GMX presents a relatively favorable factor of 12, while dYdX's ratio is less impressive at 24. For reference, Apple currently boasts a P/E of 23. It's important to keep in mind, however, that when evaluating businesses, expected future earnings are crucial. A company or protocol may have a high P/E or P/S ratio, but if it is anticipating to significantly grow earnings or sales in the future, the high ratio can be justified.

GMX is determined to increase its TVL by not only introducing new or upgrading existing products (stay tuned for the highly anticipated X4 release), but also by expanding to other blockchains. It will be fascinating to see if the protocol can maintain its advantage over dYdX, and how its oracle-based model holds against order book or AMM approaches.

Final thoughts

GMX is a formidable project that has carved out its niche in the DeFi world, proven itself to be resilient in difficult circumstances, boasts solid foundations, and has a flourishing ecosystem and grand expansion plans. Sure, there will be obstacles to overcome, particularly in terms of vertical scaling, but let's be real - creating a blue-chip protocol is no walk in the park.

From a user perspective, there are certain risks associated with holding GMX or GLP, but they are manageable, particularly given the robust and sustainable returns offered by the protocol. Does anyone still remember DeFi 2.0? I would argue it came with GMX.