dYdX: Non-custodial margin trading protocol

Quick Take

  • dYdX is an Ethereum-based non-custodial exchange protocol with margin and derivative products 
  • Open interest has spiked to $870,000 after just nine days of operation, with ETH long/short ratio currently at 13:1 
  • While the emergence of dYdX presents a material step forward in the maturation of open finance market structure, the exchange protocol has several obstacles to overcome before it can compete with its centralized counterparts 

This free preview of The Block Genesis is offered to our loyal readers as a representation of the highly valuable research and journalism our Genesis members receive daily. If you’d like to receive all Genesis content on our site and via daily newsletter, join today.

Earlier this month dYdX, an Ethereum-based non-custodial exchange, opened its revamped platform’s doors to the general public.

dYdX, founded by Antonio Juliano, previously an engineer at Coinbase, now boasts six team members. The exchange is advised by Fred Ehrsam, co-founder of Coinbase and leading crypto fund, Paradigm, and Reuben Bramanathan, previously Product Lead at Coinbase, among others. dYdX raised its $10 million Series A in October 2018 co-led by a16z crypto’s Chris Dixon and Polychain Capital’s Olaf Carlson-Wee.

Margin trading

dYdX sets itself apart from the crowded non-custodial exchange market with its margin product. Margin allows users to multiply the size of their positions, providing a path to potential outsized reward and permitting the execution of complex trading strategies. At present dYdX supports ETH, DAI, and as of earlier today, USDC.

Leveraged trading works as follows:

Alice has 10 ETH.  

If Alice wished to 3x leverage long ETH she would deposit her 10 ETH as collateral, use it to borrow 20 ETH worth of DAI, and then proceed to sell that DAI for ETH. The result is an open position of 30 ETH — her initial exposure of 10 ETH plus her recently acquired 20 ETH.  

If Alice opened her position when ETH was at $200 and closed at $220, she would net a $600 profit (30 * $20), minus the interest payment for her borrowed DAI and any exchange fees. Conversely, if Alice closed her position at $180 she would net a $600 loss in addition to interest and exchange fees.  

Trading with leverage is considered more risky than simply going 1x long in that users are exposed to liquidation.

Let’s return to Alice:

Alice is long 30 ETH with 3x leverage having borrowed 20 ETH. She entered her position when ETH was at $200 so the total dollar value of her position is $6,000 and the total value of assets owed is $4,000.

If ETH were to fall to $133.33, the value of her position would be worth just $4,000. As Alice still owes $4,000 worth of DAI (in addition to interest accrued over that period), she would be forced to sell her 30 ETH to pay back her debt. Of course, now that her position is only worth $4,000 Alice has to forgo her initial 10 ETH deposit, leaving her empty-handed. Similarly, if Alice were short 30 ETH, valued at $6,000, with 3x leverage, a move up to $300 would force Alice to cover, buying back her borrowed 20 ETH and leaving her short of her initial $2,000 deposit.  

If Alice had instead decided to buy 30 ETH with $6,000, a drop to $133.33 would leave her with a net loss of 33% but she would still get to hold onto her ETH, which could, theoretically, appreciate to a point where she is once again in profit.

In reality, dYdX’s margin service requires users to post 1.25x collateral and maintain a minimum collateral multiple of 1.15x. As such, in the scenario where Alice was 3x long 30 ETH, her liquidation price would be $153.33 and, when 3x short 30 ETH, $230. This overcollateralization requirement is necessary due to the illiquidity of the cryptocurrency markets. In this scenario, the extra $600 safety margin accounts for slippage and the compounding move in price that can occur following multiple simultaneous liquidations, otherwise known as short or long squeezes.

Compared to alternative lending platforms, dYdX’s 125% collateralization ratio is rather liberal: MakerDAO and Compound both initiate liquidation proceedings when collateral falls below 150% of debt value. BlockFi, a centralized lending venue, requires an initial 200% collateralization ratio and margin calls take place at 160%.

Positions on dYdX automatically expire after 28 days in order to comply with the Commodity Futures Trading Commission’s interpretation on Virtual Currency “Actual Delivery” in Retail Transactions, which notes that “any contracts or transactions in any commodity that are…offered to retail market participants on a leveraged or margined basis…are subject to the Commodity Exchange Act…unless delivery occurs within 28 days from the date of the transaction.”

Advantages over centralized alternatives

dYdX is not the first cryptocurrency exchange to offer margin trading: centralized platforms including BitMex, Kraken, and Deribit have all offered variations of leveraged products for some time. Yet dYdX distinguishes itself with three properties.

First, its non-custodial structure — all deposits are stored in Ethereum smart contracts — eliminates counterparty risk posed by exchange hacks, shutdowns, and seizures. Of course, unlike its centralized counterparts, dYdX is particularly exposed to the technical risk associated with smart contract security, although audits from Zeppelin Solutions and Bramah Systems suggest that there are no existing structural deficiencies.

Second, as a permissionless protocol, dYdX is open to any user with an internet connection and a Metamask account. Unlike centralized exchanges, dYdX does not require users to pass Know Your Customer (KYC) and Anti-Money Laundering (AML) procedures. This opens up access to users globally that could not necessarily have passed KYC, for better or for worse.

Third, as open source software, user-facing interfaces can be built on top of the underlying dYdX protocol, allowing any developer to create custom variations of margin products and, through competition, provide an optimal user experience. This could come down to interface design, margin requirements, or supported assets. Further, alternative protocols, like Augur, can seamlessly integrate the dYdX protocol to provide leverage for prediction market traders.    

Borrowing and lending

Lending and borrowing rates on dYdX are calculated on a variable basis. Like Compound, borrow rates for a particular asset rise non-linearly according to the utilization rate with a cap at 100% APR. Similarly, as borrowers are continuously paying all lenders, borrow rates will always be higher than lending rates. Reaching a utilization rate where 100% APR kicks in is unlikely as lenders will naturally step in to borrow at generous rates once the yield outperforms alternative platforms.

Arbitrageurs have ensured that dYdX lending rates have since come down in line with the Decentralized Inter-Protocol Offered Rate. Roughly $1.34 million in DAI is being supplied with a 15.71% borrow APR and 7.98% lending APR at 60.12% utilization rate while $1.28 million worth of ETH is being supplied with lending, borrow, and utilization rates at 0.02% APR, 0.51% APR, and 4.88%, respectively. There is currently total outstanding open interest of over $870,000 and the ETH/DAI long-short ratio is ~13.03. Because assets posted as collateral also earn interest, going 1x short ETH right now yields net ~7.10% APR absent any movements in price.



dYdX sources liquidity through other decentralized exchanges rather than building an order book of its own. At present, liquidity is solely sourced from eth2dai: other exchanges will be added once dYdX increases its range of offered pairs to more exotic assets.


These decentralized exchanges offer a fraction of the liquidity provided on centralized exchanges – a single $240,000 market sell order is enough to push the ETH/DAI price down by 5% — and thus hundred-plus basis point spreads and slippage is to be expected. Liquidity should improve as schemes are set up to incentivize market makers, dYdX sources from more venues, and on-chain throughput capacity expands through sharding and solutions like STARKs.

User experience

The first dYdX interface and user experience is more basic than the average exchange relayer: the interface lacks charting tools, instead opting to present a 1D, 1W, 1M price line chart; features like stop losses and limit orders are not presently available, forcing users to place costly market orders, and leverage is capped at 4x for longs and 3x for shorts. The platform does, however, offer Cross Margin, which allows a user to sell assets they do not have using their entire account balance as collateral, thereby extending the margin of safety beyond 125%.


Simply white labeling the Trading View interface would likely be well received by users although this does come with user experience complications. The protocol relies on the MakerDAO price feed for liquidations, rather than a single exchange, so displaying price activity on any given exchange, which is subject to differ from the MakerDAO oracle, may confuse unsophisticated users as to how secure their positions really are.

Liquidation process

In the future the liquidation process will be similar to that used by MakerDAO, whereby collateral assets are sold off to third party ‘keepers’ in Dutch Auctions at a 5% discount. Today, the dYdX team is handling liquidations itself, with ~$75k reportedly liquidated so far. Like Compound, dYdX has a built-in insurance pool, setting aside 15% of all interest earned through the platform in order to make lenders whole in cases where positions cannot be covered at prices that would return full principal. Use of the insurance pool is currently at the sole discretion of dYdX.

For now governance of the protocol, including the insurance fund, is controlled by a multi-signature contract with keys distributed across “reputable individuals with a vested interest in the success of dYdX.” The powers of the contract are limited to putting the protocol into a close-only mode, whereby users are prevented from opening new positions, and would likely only be invoked in a scenario where a major security bug was discovered. The whitepaper, last updated in August 2018, notes that, at some point, a DAO could be spawned to govern upgrades to the protocol, but details remain light.

Looking forward

The two most significant upgrades to the existing platform will be the addition of more assets and the ability to trade options, which themselves will be represented by erc20 tokens. The addition of options will open up a new dimension of trading opportunities for decentralized finance users — collars, straddles – and present a valuable underlying asset for advanced Set token strategies.

However, the path to success for dYdX remains relatively unclear. Liquidity is always front of mind for traders and sourcing from decentralized exchanges is unlikely to ever seriously compete with the order books of centralized exchanges — aggregate DEX volume over the past 24 hours is roughly 2% of volume processed by Binance.

This is largely a function of the Patriot Act’s strict KYC/AML legislation, requiring institutional market makers to know their counterparties, but also a structural problem — base layer chains are currently severely rate-limited. Front running from miners, which complicates execution guarantees and prices, is also a severe disadvantage, as recently explored by researcher Phil Daian.

Lending rates on decentralized protocols are also unlikely to compete with centralized alternatives, largely for the same reasons: BlockFi, for example, offers retail customers a fixed 11.25% rate on USD loans for 200% collateralized loans, over 4% less than rates on dYdX.  

Further, the Ethereum blockchain’s transparent nature opens up margin traders to some additional risk: a sophisticated user could develop a script that builds a log of open positions at different price levels and sizes. This could be sold to a well-capitalized entity, who could then proceed to stop-hunt orders. The equivalent would be BitMex opening up its trade history and liquidation levels for all to see.

It also remains unclear how the dYdX team can capitalize on their invention and generate a sustainable business model — indeed, this difficult path to value accrual and moat building applies to the wider open finance industry, which, almost by definition, is acutely subject to low entry barriers. A fee on each trade could be viably introduced, or some percentage of interest could be reserved, but the dYdX team would have to ensure that their interface provides the best user experience possible: competition from other teams building on top of their protocol will likely be rampant.  

Nevertheless, the emergence of dYdX is a material step forward in the maturation of the open finance market structure. The ability to seamlessly short tokens will catalyze price discovery, hopefully putting an end to the high correlations we see in today’s crypto market. The ability to trade in a non-custodial fashion will prevent the kinds of exchange hacks witnessed by Binance earlier this week. Liquidity issues posed by the absence of institutional participation can potentially be solved through 0x’s Permissioned Liquidity Pools.  And the emergence of options will provide a powerful arsenal of strategies for both sophisticated and passive investors through structured product platforms like Set Protocol.