Derivatives — particularly perpetual futures (“perps”) — are undoubtedly the largest vertical in crypto, comprising the bulk of total crypto trading volume.
Virtually all perps trading occurs on centralized exchanges. The irony that most retail and institutional traders alike trade decentralized assets on centralized servers is not lost to many in the community. Over the past few years, decentralized exchanges (DEXs) built on top of smart contracts on blockchains emerged. For the crypto-savvy, DEXs like dYdX and GMX have gained considerable attention and stand as viable alternatives to centralized incumbents. Still, volumes and active users of these decentralized alternatives are still only a fraction of their centralized peers.
Our team believes that decentralized derivatives will be a secular growth vertical that will slowly but surely take share from centralized exchanges. We explain our thesis in this report, focusing on perps because it makes up the majority of all derivatives trading volume. First, we examine the overall market structure for both centralized and decentralized perps. Then, we explore the merits and drawbacks of various DEX designs. In particular, we look at the viability of decentralized perps that rely on aggregated liquidity pools.
In a subsequent report, we will build upon points in this report to evaluate three upgrades to existing DEXs: GMX v2, Synthetix v3, and dYdX v4.
Derivatives refer to a financial instrument that derives its value from an underlying asset. They exist in all markets. Contracts can track anything from S&P 500 and gold to orange juice and the weather.
Because derivatives are merely contracts that track value, they can be constructed to have all sorts of payoffs. Most common are futures, which have linear payoffs and are akin to buying the underlying itself. Perps, the focus of this report, have a linear payoff at 1x leverage.
Most futures in traditional markets are dated futures. They expire at a certain date, and at expiration date, the price of the futures contract and the price of the underlying asset (called the spot price) converge to be equivalent. Before expiration, futures prices and spot prices can vary, sometimes significantly.
In crypto, however, perpetual futures are the most popular derivatives product. As the name suggests, these contracts exist in perpetuity. The price of these futures is anchored to the underlying price via a funding rate mechanism. Periodically, typically every one or eight hours, money flows from one side of the contract to another.
For example, if there are more longs than shorts on BTC-PERP, the price of the perp will be higher than the underlying BTC index price. To adjust for this, those who are long the contract have to pay a fee to those who are short the contract, incentivizing the perpetual future price to match the index price. Find a simplified, thorough guide on perpetual swaps in this explainer by Paradigm.
Likely the reason perpetual futures became the preferred product is because crypto liquidity is scarce. Dated futures fragment liquidity across multiple expiries, whereas perpetuals consolidate all available liquidity for an asset into one instrument. Related to this, perps are relatively more accessible to retail.
Derivatives serve two purposes: hedging and speculation. But it would be more helpful to think about why they exist at all (i.e., why participants do not simply hedge or speculate in the spot market instead).
There are two key reasons: inaccessibility to the underlying and leverage. A high-friction spot market creates the demand for liquid derivatives markets. For instance, it is a lot easier to sell wheat futures than it is to take a short position in the underlying wheat market (which would presumably involve borrowing a lot of physical wheat). Furthermore, transacting in derivatives is more capital efficient as traders can use leverage, putting down only a small portion of their capital to achieve a notional value that could be sometimes 20x, 50x, or even 200x their margin. With only $100K, a trader can gain exposure to $5M of an asset with 50x leverage.
It is more due to the latter reason that crypto perps grew in market share over the last five years. When BitMEX first created the Bitcoin perpetual swap, it offered up to 100x leverage, meaning that traders could lose their entire capital in 1% moves. Over the past few years, perps pairs also have expanded to a wide range of different crypto assets, allowing users to also hedge/speculate in harder-to-access digital assets as well.
Perpetual futures are the largest traded instrument in crypto. To provide a sense of scale, futures volume has comprised ~75% of total trading volume for Binance since 2021. Other non-US crypto exchanges have similar or higher proportions. The story is different for the US, where regulations prevent exchanges like Coinbase and Kraken from having a robust derivatives market. It is no wonder that Coinbase opened an offshore crypto derivatives exchange just a few months ago.
However, virtually all perps trade volume occurs on centralized exchanges. To give a sense of how much centralized volumes dominate decentralized ones, consider that volumes for spot DEXs are often 10–20% those of CEXs. In fact, Uniswap’s monthly trading volume exceeded Coinbase’s for four consecutive months this year. In contrast, DEX penetration is only 1–3% for perps.
Probably the main reason for this disparity between the spot and perps verticals is the greater difficulty in creating a perps DEX that is on par with a CEX. While it is relatively simple to create an acceptable spot DEX (variations of x*y=k), it is orders of magnitude more challenging to do the same for perps due to multiple moving parts — such as managing margin requirements, cross-margining, funding rates, leverage ratios, price oracles, insurance funds, and more.
Moreover, because perps are cash cows for centralized exchanges, competition is intense. Not only do DEXs have to catch up to CEXs, but they also have to keep up with how fast multiple centralized entities innovate and improve on perps.
Yet, the prize is immense. Exchanges across all markets exhibit a power law, where the winner takes predominant market share and a select top few exchanges take up almost all the market. This is primarily due to network effects (liquidity begets liquidity) and scale economies (larger exchanges can more efficiently re-invest in operations).
For centralized exchanges, the leading exchange typically takes up 60–80% of total perps volumes. Historically, BitMex dominated perps volume because they first invented the product. However, a period of fierce competition ensued from OKX, Binance, ByBit, and others starting in 2019. Since 2021, the industry has gradually consolidated, especially after FTX’s collapse in 2022. Today, Binance makes up ~60% of total perps volume. The top three exchanges have ~90% market share.
We believe derivatives DEX will share a similar market structure where the leading exchange holds the majority market share. If this assumption holds correct, the potential prize of the market winner is well worth over $25B.
Below, we illustrate how a hypothetical leading DEX could grow its valuation from $500M to over $15B if executed correctly. And if DEX volumes ever come close to parity, or even overtake, CEX volumes, these estimates could prove to be very conservative.
In order to validate our thesis, we need to understand why traders will prefer DEXs to CEXs in the first place. Below, we outline a selection of key factors that traders often prioritize when choosing an exchange regardless of DEX or CEX. While this list doesn’t cover everything, it’s a good starting point in evaluating how DEXs can win.
Although all factors are important, liquidity is the most critical aspect of an exchange. Without it, an exchange isn’t valuable to buyers, sellers, or market makers. Most exchanges stagnate or fail due to lack of liquidity.
All other factors arguably serve to improve liquidity on the exchange. For example, cross-margin functionality significantly lowers capital costs for market makers, helping to improve liquidity. Robust infrastructure allows professional market makers to manage their risk when providing bids and asks. Traders will also prefer performant and resilient exchanges with safeguards to manage black swan events.
Decentralized perpetual exchanges often attempt to win via touting the lack of custodianship / KYC while offering token incentives and negligible trading fees. dYdX leveraged this strategy with great success in 2021, rapidly gaining trading volumes and users, cementing itself as a top decentralized trading venue. Buoyed by dYdX’s ascendance, newer exchanges like ApeX, Level Finance, and Kwenta have adopted a similar playbook.
However, this inevitably leads to traders finding ways to game the incentive program, typically through wash-trading to farm token rewards. Most of these traders churn after the incentivization program ends. Furthermore, these traders often do not add meaningful liquidity to the exchange — they often shoulder little market risk and merely trade against themselves (or other farmers).
These case studies suggest that non-custodial / no-KYC factors are not enough to take meaningful flow from CEXs. Thus, DEXs should think carefully about incentive programs — they are useful complements to kickstarting liquidity but poor in attracting organic flow.
DEXs will also probably not win on performance. DEXs will always be less performant than CEXs 99% of the time. CEXs have advantages in latency, throughput, finality, etc. due to running on centralized servers. Although DEXs can theoretically be more resilient during black swan events.
Instead, we believe perp DEXs will gain penetration by competing where CEXs cannot, similar to how Uniswap and other spot DEXs gained prominence. Looking at the list above, we list a few examples of value propositions that only DEXs can offer:
Composability. By having parts or all of the DEX infrastructure on-chain, other DeFi projects can plug into a perps DEX to create new parts. This is already playing out with some of the larger on-chain perps DEX, such as GMX and Synthetix. For instance, there are multiple projects on Arbitrum that compose with either GMX’s liquidity pool (GLP) or its perps exchange (such as Rage Trade, Rysk, and more). As another example, Lyra leverages Synthetix’s perps engine to create a delta-hedged options market-making vault.
We believe composability is primarily where DEXs will win. An ecosystem of motivated entrepreneurs building and interacting with each other in a composable and permissionless DeFi system will gradually but inevitably outpace a team of in-house employees of a centralized exchange.
Decentralization. Decentralization is not merely a virtue — they also improve the unit economics of DEXs compared to CEXs. DEXs outsource many key functions, improving the cost structure that allows more capital for reinvestment. For instance, decentralized exchanges do not need to invest in custody of assets, nor do they need to hire lawyers to determine which assets to list. Settlement of transactions is also typically outsourced to the blockchain layer. And, at the extremes, decentralized projects also do not need to geoblock or KYC users, although this may be only a fleeting advantage due to regulatory inertia.
Decentralization via token distribution also incentivizes a grassroots community to market and build upon the project. GMX’s “Blueberry Club” and dYdX “Hedgies” members actively promote their respective platforms across Twitter. Due to token ownership, members also help build dashboards, analytics, and even other projects on top of perp DEXs.
Overall, DEXs will most likely win by adopting strategies and operating models that CEXs structurally cannot mimic.
There are three dominant DEX designs:
Central limit orderbook (CLOB) — an orderbook that collates all buy and sell orders for an asset, in which trades occur where the buy and sell prices match.
Multi-asset pool — a pool that consists of multiple assets, including crypto and stablecoins, that is used to facilitate trading by acting as a counterparty.
Single-asset pool — a pool that consists of only one asset, often either a stablecoin like USDC, that is used to facilitate trading by acting as a counterparty.
CLOBs are relatively straightforward. They are a proven model for traditional exchanges run by centralized entities. The key difficulty in porting this model to blockchains is the higher cost of compute, bandwidth, and storage of blockchains. Currently, most teams resort to managing computationally expensive parts off-chain.
The latter two models, on the other hand, are crypto-native designs that emerged due to immature on-chain markets and the high transaction costs inherent in blockchains. Projects that use this model rely on a pool of assets and a price oracle to execute trades against this asset. While a simple idea, they present a whole set of other challenges.
Because pools are newer and more unproven designs, we dedicate more time to exploring the merits and drawbacks of pooled designs.
Does the House Always Win?
In pool designs, the gain of traders often comes at the cost of liquidity providers, and vice versa. Therefore, their sustainability and success hinge on validating the thesis that “the house always wins.”
To evaluate the merits of the thesis, we can first look at the history of these crypto products themselves. GMX’s statistics dashboard shows that, since August 2021, traders on the platform have been net unprofitable. However, extended bouts of trader profitability occur, primarily when prices are trending upwards. Likewise, Synthetix’s Perp v2 shows cumulative net trader loss with spans of positive PnL.
However, the sample size for these data is limited. We can further draw on statistics from traditional markets to verify the hypothesis. Perhaps the closest comparable to crypto perps are CFDs (Contracts for Difference) because they are 1) derivatives on underlying assets, 2) allow for extreme leverage (some up to 500x!), and 3) are retail-driven products.
The data is supportive. In the US, firms typically disclose ~70% of client accounts to be unprofitable on a quarterly basis. In Europe, regulation by the European Securities and Markets Authority (ESMA) requires brokers to report profitability rates of retail clients. Across all brokers, ~74% of clients are unprofitable. In France, a study found that the overwhelming majority of retail investors (89%) experience trading losses over a four-year period. Similarly, a Spanish report found that 75–85% of clients suffer losses.
Given that most crypto perps allow for comparably high leverage on an asset class that have orders of magnitude higher volatility, we could reasonably assume that similar statistics apply to DEXs.
Beating the House
But caveats apply. The main risks to the “house always wins” thesis are adverse selection and toxic flow.
Adverse selection is when a venue attracts only sharp traders. At the extreme, unprofitable traders no longer trade on the exchange. This could occur if all unprofitable traders churn (their bankroll goes to zero) and no new chumps are introduced to the platform. This risk is more pronounced for market laggards. If no new traders onboard to the platform, the DEX’s LPs risk being counterparties to sharp traders who opportunistically trade against the LPs only when it is profitable to do so.
Related to the above is the risk of toxic flow. Toxic flow is orders that have a high probability of short-term trading profit due to information asymmetry. The most common source of toxic flow for pool DEX designs has been from traders’ front-running price oracles. For instance, if a trader sees that the ETH price on Coinbase has suddenly moved from $1,850 to $1,900, the trader can attempt to trade on an oracle-reliant DEX at the stale price of $1,850, wait for the price oracle to update ETH’s price to $1,900, then sell at a profit. This is largely a solved problem with new pull-based oracle designs but vulnerabilities may still exist.
Toxic flow can come from many other sources too. A savvy trader could attempt to capitalize on low-liquidity events by buying a large position on a perps DEX, then manipulating the price of the underlying asset to move the perp position to profit. Or, traders well-connected to multiple venues could also statistically predict the likelihood of price movements in the short-term, and use DEX perps as a venue to profit.
Another factor in pool DEX designs is what we term the “long crypto” phenomenon — most participants in crypto tend to be net long. Single-asset pools denominated in USDC are particularly vulnerable to this phenomenon as LPs will take on the other side of the trade (net short). So in a trending bull market, LPs are highly likely to experience large losses. This could spark a “death spiral” where LPs pull liquidity, reducing available exchange liquidity, deterring users to trade on the platform, thus sparking LPs to withdraw liquidity, and so on. In contrast, multi-asset pools are easier to manage as the pool itself is naturally long crypto as well.
Key to managing this is incentivizing traders to take the other side of the trade and/or increasing the costs of crowded trades. For example, Synthetix’s dynamic funding rate has helped to significantly mitigate this risk.
Pool-based DEXs certainly have their place in the perps vertical as they offer attractive features for both the demand and supply side of an exchange. On the supply side, they somewhat level the playing field by effectively allowing liquidity providers to outsource price discovery to other markets. Both retail and institutional LPs alike provide the same quote to traders. On the demand side, they can provide competitive price execution for traders.
Nonetheless, the terminal value of these exchanges is likely capped because price discovery always occurs on other venues. At the extreme, it is unfeasible for pool-based exchanges to have over 50% market share. Furthermore, their growth is bottlenecked by oracle support and LP capital inefficiency.
Hence, while they serve important roles as alternative trading venues with strong growth potential, we believe the leading derivatives exchange will be one that can enable endogenous price discovery, which is almost certainly going to be a CLOB-style exchange. CLOBs offer traders the best capital efficiency and most effective way to express preferences. Their traction has been somewhat limited by current blockchain infrastructure, but we see several upcoming innovations in the near term that will improve CLOB DEXs. In the meantime, hybrid CLOBs — where order matching is performed off-chain, but settlement is on-chain — is an appropriate trade-off.
Compared to CEXs, the current DEX market structure is immature and fragmented. dYdX often leads in trading volume, as it is a first mover in perps DEX and has onboarded multiple market makers. However, there are multiple periods where SNX has higher volumes — the latest spike coinciding with a trading incentives program. GMX, the third place for weekly volumes, sees steady growth due to grassroots community support.
Value accrual is a different story. GMX and SNX tokenholders can participate in the upside of their respective exchanges, staking tokens to earn a portion of platform fees. On the other hand, while dYdX earns revenues through fees, tokenholders do not participate for now.
And in terms of actual active users, dYdX and GMX tend to lead. Nonetheless, all DEX metrics are negligible compared to those of CEXs.
We reiterate our belief that the decentralized derivatives sector — and particularly decentralized perps — is a secular grower for the next few cycles. For Part 2, we build upon this report to look at the various live and upcoming DEXs, and dive deeper into the designs for three particularly exciting upgrades: GMX v2, Synthetix v3, and dYdX v4.
Preview for Part 2
If you have read this far and have questions / thoughts / disagreements — please reach out.
Steven Shi is an Investment Partner for Amber Group’s Eco Fund, the company’s early-stage crypto venture fund. He has over five years of experience in hedge funds, private equity, and investment banking firms.
The information contained in this post (the “Information”) has been prepared solely for informational purposes, is in summary form, and does not purport to be complete. The Information is not, and is not intended to be, an offer to sell, or a solicitation of an offer to purchase, any securities. The Information does not provide and should not be treated as giving investment advice. The Information does not take into account specific investment objectives, financial situation or the particular needs of any prospective investor. No representation or warranty is made, expressed or implied, with respect to the fairness, correctness, accuracy, reasonableness or completeness of the Information. We do not undertake to update the Information. It should not be regarded by prospective investors as a substitute for the exercise of their own judgment or research. Prospective investors should consult with their own legal, regulatory, tax, business, investment, financial and accounting advisers to the extent that they deem it necessary, and make any investment decisions based upon their own judgment and advice from such advisers as they deem necessary and not upon any view expressed herein.