dYdX and StarkWare shared the reasons why dYdX chose StarkEx as the expansion solution, and compared various Layer 2 solutions.
Organize: dYdX
On April 22, dYdX invited Ohad Barta of StarkWare and Brendan Chou of dYdX to participate in an AMA with the theme of “Building a Decentralized Perpetual Contract Exchange on StarkWare”. During the event, Ohad and Brendan shared the following:
- Their background and experience in entering the crypto industry
- Why should dYdX migrate to Layer 2
- Why dYdX chose StarkWare as its partner
- How StarkEx works and the different data availability models
- StarkNet and future plans and visions
- The nuances of Stark Key and how it is used in the Layer 2 system
- Oracle performance on Layer 2
- Comparison of ZK Rollup and other Layer 2 technologies (such as Optimistic Rollups/ ZkSync)
- The relationship between spot and derivative products
The following is an edited interview transcript:
David: Thank you for your participation. I am David Gogel, a member of the dYdX growth team. I am very happy to invite Ohad from StarkWare and Brendan from dYdX today. We will discuss some issues from the community and the technical implementation of dYdX StarkEx’s cross-margin perpetual contract supported by StarkWare. Brendan, Ohad, can you share your background and how you joined dYdX and StarkWare, respectively?
Brendan: My profession is a software engineer. I graduated from Princeton University and worked for Bloomberg, Google and other large software companies, where I learned some work skills, how to program, and how large companies work. At the same time I am also looking for something different. I am Antonio (the founder of dYdX) as a roommate in college. He went to Coinbase as soon as he graduated from college, so I always asked him to explain to me what Bitcoin is and how it works, but I never fully understood it. I started to research more things at the peak of 2017.
I know he is starting his own company. We gather with all our former roommates to celebrate the New Year every year. We talked during this time and that was when I decided to join. I was in New York at the time, and I realized that if I could take a step and engage in this crazy encryption work, it would definitely benefit me. This is my experience in entering the industry.
David: Ohad, how did you join StarkWare?
Ohad: Eli Ben-Sasson, one of StarkWare’s co-founders, and I have known each other since 2011. I wrote part of the first version of PCP Prover and Verifier for him, which are the old versions of STARKS and SNARK used today. Then I spent six years gaining programming experience in the Israeli intelligence service. Later, I got a master’s degree in cryptography from Professor Yuval Ishai, and we had a cryptography thesis last year. After that, I contacted Eli and asked him what’s going on with StarkWare. After seeing the proof system become so mainstream in the past six years, I decided to join Eli.
David: A few weeks ago, dYdX publicly released a cross-margin perpetual contract on Layer 2. Brendan, can you provide more background on why dYdX decided to move to Layer2, and the advantages of the new Layer2 products over Layer1 products?
Brendan: There are many reasons. The most important is the order execution part. Previously, this was an Ethereum transaction to settle every transaction on the chain. This also brings some challenges. First of all, it is not instant and can be rushed away. This brings some uncertainty to everyone. You have to wait for the block to be mined before your transaction can pass.
Moreover, our own transactions require a lot of gas fees. This is a waste of our money, which is unsustainable for us. We have to make users bear things like higher fees or higher minimum transaction amounts. As the scale of Ethereum expands or the number of users expands, this is not good for anyone. So I think this is our main motivation, the order execution part.
Another problem we saw in Layer 1 was the speed at which the oracle price was obtained on the chain. When the price changes sharply and the gas cost rises, the price of the oracle is reported to Layer 1 much slower, which brings some additional risks to users. Therefore, we must punish liquidated users by providing a lower leverage ratio and a more aggressive liquidation system. Similarly, it costs a lot of Gas, is slow, and contributes less to the security of the entire system. So I think this is the main factor we are concerned about, we will migrate to Layer 2 to get more instant certainty, which will bring a better experience and lower costs.
David: Why did you choose StarkWare as a partner to develop the Layer 2 system?
Brendan: What we want is rapid development, and we hope to release the product in the next few months. So we want to eliminate more risks and cooperate with those who have been on the mainnet for a while. StarkEx has been cooperating with DeversiFi and has achieved great success, and has gained a good reputation from the people we have spoken to. So I think these are the main influencing factors.
David: Ohad, can you briefly talk about what StarkEx is? And how does it work?
Ohad (StarkWare): StarkEx is an expansion solution provided by StarkWare for applications on Ethereum. It takes a sequence from the sequencer (dYdX in this case), runs them internally, and ensures that everything is checked out and meaningful. Then, it moves the transaction to the Cairo program. Basically, the Cairo program can represent any statement (it has Turing completeness). The Cairo compiler will compile the Cairo program, and then our prover will convert it into a STARK proof. Then, we send the proof on this chain to the verifier for verification. If the verifier accepts the proof, the proof is legal. Therefore, we can achieve accurate and instant settlement.
David: Can you describe the different models of StarkEx and their trade-offs?
Ohad (StarkWare): For Validium, we don’t publish them by status, because the cost of doing so is very high: even if it only costs a small part of the actual execution of the transaction, it is still non-negligible and changes with changes The number is increasing linearly. Instead, we would say: “Well, we have this trustworthy server committee.” This is the approach taken by DeversiFi and Immutable, which have several respectable committee members. In this method, as a condition of signing the batch, the committee members must prove that they all know all the relevant data that the user may use when the operator fails. In ZK-Rollup, we will publish a new balance on the chain every time we update. Therefore, users have more trust in the system, but the cost is also higher.
David: Brendan, dYdX’s StarkEx implementation is ZK-Rollup, not Validium. Can you describe the thinking process of why ZK-Rollup is a better model for dYdX?
Brendan: Actually, we don’t have any particularly strong ideas about this. I think Ohad describes the trade-off very well there. It just makes the data available to everyone, but the price of publishing is that it requires more gas to do so. Therefore, we decided to temporarily keep it on the chain and aggregate it, because we feel that we can afford gas. At present, we have not passed the cost of Gas to users. So there is no gas fee when trading. Assuming that the price of ETH continues to rise or the network usage rate increases, we can delete the part that publishes all data on the chain in the future, and then maybe ask some additional committee members to verify that the data is stored in multiple different locations. Used by multiple different participants, so users will not lose track of the actual current balance on the chain. As Ohad said, this is important. If they need to basically prove their position and exit the system.
But this is only the way forward, because if the cost of doing so is too high, then, as I mentioned, if these costs must eventually be passed on to users, communities and other users, this is where we can put more data Off-chain, saves more gas fees for everyone.
David: With the transition to Layer 2, people can see less data in the block browser. Ohad, can you describe the data available for users to view on Layer1?
Ohad (StarkWare): The data that can be viewed on Layer1 is the balance of all users, including synthetic assets and capital index: This information can know how much funds we really have when the last batch is received. The transaction itself is not released on the chain, which is not only to save costs, but also to keep the transaction strategy private.
David: As a user, how can I believe that the transaction data kept outside of Layer 1 is actually correct? Is there any mechanism to ensure the validity of everything that happens off-chain?
Ohad (StarkWare): Before discussing StarkEx, I said that all of these are based on the Cairo program, which is the center of everything. So what happens is that the Cairo program will get the order of the transaction. In addition to the commitment to the output of the new state, it also adds all position balances that were changed during the batch. And this output is linked to the certificate. If you cannot show me the transaction through the data on the chain, the proof will not be accepted.
David: This answer is really helpful! Ohad, can you provide an update on the status of StarkNet and its vision?
Ohad (StarkWare): Maybe I will start with StarkNet, because so far we have only discussed it on StarkEx. StarkEx is cool. It does help you now, but it is not enough in the long run. In the long run, the reason why it is not good enough is that StarkEx currently requires operators to operate, which is disadvantageous from a centralized point of view, technical and regulatory point of view. StarkNet means that all Cairo programs and Prover/Verifier are placed in the center of StarkEx, instead of working with a sequencer that will only build a complete side chain, where each side chain Each block will be proved by comparing it with Cairo, and then the proof will be as we proved today, even using the same verification procedure as today.
Of course, this will take some time, so we proceed in steps. Therefore, at the end of May, we will publish a demo on StarkNet’s Ropsten that will allow the writing of custom contracts. Take the AMM contract as an example, but it does not yet allow interaction from Layer1 to Layer2. Then, a few months from there, we will start to expand StarkNet to allow interaction between contracts and Layer1 and Layer2.
Then, the next thing we need to do is to decentralize the operator. Therefore, we will indeed have a complete side chain of miners and everything, not just a sequencer that sorts all smart contracts submitted by users.
David: Brendan, as part of the process of getting started on dYdX, users connect to their Ethereum wallets and conduct transactions. But for the first time, the user must create a Stark Key. Can you describe what Stark Key is and how it is used on dYdX?
Brendan: Basically, it is similar to an Ethereum private key. It is just an encryption method that verifies that everything you sign comes from your private key. You must use different private keys or different elliptic curves to sign things that are compatible with the STARK system. I don’t have too many technical details. Ohad should know this better than me. But in essence, this is just another key for you to keep track of everything.
We don’t want users to have to remember another complete private key. They are used to protecting their Ethereum keys. We don’t want them to need to take care of another thing. If they lose it, it’s bad. We came up with a system that allows them to use their existing keys to derive it. This is basically obtained from hashing their fixed signature, as long as they have the Ethereum private key. Therefore, as long as they have their own Ethereum private key, they will never lose their Stark private key.
This is really useful because, as I mentioned, they don’t have to keep two things. And similar to the way you can’t really forge someone’s Ethereum signature or find a hash collision, their private Stark key is also very secure.
David: Do I have to protect my Stark key like a private key? If someone can get my Stark key, can they get my funds?
Brendan: This is a very complicated question. Yes, you should definitely keep it secret. I’m just thinking about the worst thing people can do, and I think in this case, they might sign an order on your behalf. However, if this is all they have, even if they withdraw your funds to Layer 1, they cannot remove these funds from the contract. They still need your Ethereum key to do this. In addition, on our side, we also have API keys and secrets to prevent people from doing these things without explicit permission.
Similarly, if only your Stark key is compromised, in terms of dYdX, unless you also have an API key, we still do not accept requests. API keys are another small set of private keys, which can also be derived from your Ethereum private key. Therefore, if you own the Ethereum private keys, you can at least own those too. These three keys are all derived from the Ethereum private key. So if someone takes your Stark private key, it’s not good, but there are still many security mechanisms to prevent someone from causing damage there.
David: Ohad, can you describe which hardware or software wallets currently support Stark Keys?
Ohad (StarkWare): Currently, the Ledger wallet natively supports Stark Keys and StarkWare applications, as well as several other wallets. I believe that Authereum supports it, although it has recently been deprecated. For any other wallet, we have a JavaScript library that requires the wallet to be signed with an Ethereum key, and the Stark key is derived from the signature, as Brendan said. This is how MetaMask and all common wallets currently work. We plan to add more wallets to support Stark Keys natively. This is the goal we have been pursuing. As more and more projects like yours go on, I think this will happen soon.
David: After the flash crash of the cryptocurrency market last weekend, a lot of discussions were triggered. StarkWare published a blog post about the price of the oracle on Layer 2, which seems to be a big problem on other DEXs. Brendan, how does dYdX’s StarkEx implementation use Oracle’s price? How is it different from Layer1Oracle?
Brendan: So now we only use Chainlink oracles. We are also communicating with several other oracle projects to see if some cooperation is needed. As I said, dYdX is seeking rapid development, and what is commendable is that Chainlink can provide us with a large number of different trading pairs and supports Layer 2 keys or Stark keys.
Currently, we only work with one oracle project, and they send us encrypted and signed prices. We can use encryption to tell them that they are from the signer or someone whose private key we trust. Then, we sum them up and use their median value to become the price of the Layer 2 oracle. All prices are stamped with things like timestamps and asset names. Put them together and get the median of a certain arbitration. This is how we perform operations on Layer 2. This is very similar to the pattern of many oracle projects that everyone has seen on the Layer1 network.
In addition to not requiring every oracle to send transactions, our approach is similar to aggregating all signatures in batches without the need to package all transactions on the Layer1 network. They can directly enter the Layer 2 block and finally go on the chain when the entire block is packaged and released.
David: Ohad, can you describe what happened over the weekend and how the Layer 2 oracle helped dYdX maintain its pricing?
Ohad (StarkWare): Last weekend, the price of Ethereum dropped by 20% in a short period of time. The biggest advantage of dYdX over other decentralized exchanges is that we continuously obtain all prices from the Chainlink oracle, and can react to price changes in real time, without having to wait for the transaction to be accepted on the chain. After we obtain the price from Chainlink and use it, we can send subsequent transactions through that price again without waiting for what happens off-chain.
This is why dYdX can run normally. We have conducted more than 1,000 clearings in one hour, which is impossible on the Layer1 network, because many of our competitors are also sending transactions at this time. Therefore, this is why a smaller delay can improve system stability.
David: Ohad, can you describe the current role of the ZK verifier? What are the future plans for decentralization of this technology stack?
Ohad (StarkWare): Of course, currently StarkWare is running a verification program. The verification program obtains the Cairo program from the StarkEx component, generates a certificate, and then sends it on the chain. I am sure that in StarkNet, the proof will be decentralized in some way, but since this is the last step that StarkNet will complete next year, I can’t really promise our approach.
David: From the user’s point of view, what is the risk of the demonstrator in the centralized zero-knowledge proof? Can you review my transaction? Is there any downtime for this proof? How does this affect the stability of the entire system?
Ohad (StarkWare): Verification should be online in real time. We use the best services provided by Amazon, and these services are running around the clock. Even if something happens, they will be down for a short period of time without affecting the system.
As for the question of review, this has nothing to do with the fact that the demonstrators are too concentrated, but to the fact that the ranking itself is concentrated. Therefore, the transaction may be subject to scrutiny. However, user withdrawals cannot be reviewed.
If you send a transaction through the system and see that the system ignores you for some reason, you can submit a Layer1 request and withdraw, or even trade your synthetic assets, so that you don’t have to pay a maintenance deposit for the synthetic assets you own. Using a series of such Layer1 transaction processing, you can completely shut down the system. Moreover, if the operator chooses to also review those requests supported by the Layer1 certification, then your user review system can now go to the Layer1 smart contract and tell him: “Look, my smart contract, I should withdraw the request a few days ago , But I’m still under scrutiny.” Then: Everything is closed, of course, we have never been interested in doing such a thing.
David: Brendan, how does the dYdX team consider maintaining non-custodial transactions on the Layer2 network? Compared with the Layer1 network, what trade-offs must the team make?
Brendan: As mentioned by Ohad, now we are basically choosing the order of transactions, which is slightly different from the Layer1 network. Usually in the Layer1 network, the miners choose the person with the highest gas fee. It’s a bit like we are the miners of each block. But as Ohad mentioned, there needs to be certain protective measures so that people can withdraw funds anyway, so you can’t review all transactions.
People can agree to make a transaction, and then send it back to the Layer1 network, and we will force it into the Layer2 block. If it is to submit a withdrawal request, we must do the same. Therefore, as long as people can get their money back, we will encounter the same problems as the Layer1 network, which is still non-custodial in the sense that we can’t steal funds, but for now, because everything is going into a more centralized Matching engine, we can choose not to allow certain traders on the system.
Therefore, we will continue to work on further thinking in terms of strategy and become more decentralized, especially in order execution. I think that in the case of very tricky order execution, non-custodial nature will be far ahead and more decentralized.
David: Do you have any preliminary thoughts on a decentralized order book and matching engine?
Brendan: This is a difficult question, so I won’t introduce too much here. Usually, even in traditional finance, you will see many different types of trading systems. What we are currently using is the order strategy. I know that 0x and some other projects are using more inquiry systems or request quotation systems. You basically ask someone to give you a quotation and then agree to it. This is a bit different from how orders work.
Therefore, it may be necessary to study this type of model more, and even Uniswap can do more AMM-type models to open up the situation. I think there are many different ways. I don’t have anything particularly specific right now.
User questions
User: My question specifically concerns the market and how to add different assets. If I remember correctly, dYdX has Link, Uniswap, Solana and Sushi trading pairs. In the future, will dYdX find ways to involve the community in the choice of adding new projects? Or continue to be done internally?
Brendan: This is definitely something we want the community to participate in, so that everyone has more say. As an exchange, the most important goal is to let people trade what they want to trade. For us, it is important to involve the community so that people have a greater say in what they want to see. So this is definitely our top priority. I think that maybe in the next few months, in 2021, we hope to give people more say in this regard.
David: I want to add that another benefit of migrating to Layer 2 is that we can add more trading pairs faster. In the past 2 weeks, we have added 4 trading pairs. Our current goal is to add one or two trading pairs per week for the rest of the year. By the end of this year, there are 30 to 50 perpetual contract trading pairs. The decision-making process surrounding it is very focused on listing the tokens that our users want to trade. What we assume is that users want to trade top DeFi tokens and top Layer1 tokens. We always value the opinions of the community. If you have a particular trading pair you like, please feel free to share it with the administrator so as to provide us with reference.
Brendan: As for the other part of your question, it mainly depends on who is making the decision now. At present, it is basically decided by the dYdX team.
David: There is a lot of discussion about different Layer 2 solutions. Recently zkSync issued a statement. Their vision for zkPorter is that I think in some respects it is very similar to StarkWare’s Validium model. Ohad, what do you think of zkSync? What is the vision of zkSync? How is it different from StarkWare’s goals?
Ohad (StarkWare): First of all, in order to make the zkPorter part clear, zkPorter is the idea of combining Validium with the ZK-Rollups method we discussed earlier. We called it “Volition” on StarkWare, and we even discussed it as early as June 2020. We are very happy to see that others in this field are adopting some of our ideas and will soon be live on many mainnets.
In general, I think from a technical point of view, zkSync is somewhat close to what we do. They are another company that provides proof of universal validity and scalability, at least starting from the next version, because Currently they only have transmission.
User: Sometimes I find that my profit margin sometimes exceeds 100%. If the price is too high, I will not be able to close the position, because the closing operation will cause the margin to still exceed 100%. Is this intentional or a mistake?
Brendan: What we ultimately want is that as long as you are closing the position, it should allow it to pass. There are multiple different verification steps throughout the process to allow this to happen. We basically don’t want people to open positions that exceed this limit. Even in some cases, if we allow this, it may prove It becomes invalid by itself because we allow someone to assume an invalid position.
But I think we may be too sensitive to this now, and we need to stop in some places in the verification system. So I think we are asking people to go back to the initial margin level now, but in the end we do want to allow this. I think a few people have sent us messages recently, so I think we might give priority to this matter. We do want to allow this to happen, because especially I think people with large positions may not want to close them all to fall below 100. Thank you for raising this question.
David: Ohad, can you transfer funds between different StarkWare solutions? For example, switch from DeversiFi to dYdX and back?
Ohad (StarkWare): Assuming you mean some kind of untrusted or proven transfer, we don’t support it yet. We intend to use StarkNet to support it, and StarkNet will provide application communication between contacts from the end of this year. In this way, these transactions will be easily realized.
David: At some point, will users be able to transfer funds from one Layer 2 application to another Layer 2 application?
Ohad (StarkWare): If you consider the Layer 2 application that StarkWare runs, you can be sure. I think in the long run, we are also willing to cooperate with other Layer 2 solutions, but this is a long time frame.
David: Ohad, can you describe some of the main trade-offs between zk rollup and Optimistic rollup at a high level?
Ohad (StarkWare): Let’s briefly introduce Optimistic Rollups, because so far we have only discussed zk rollup. In “Optimistic Rollup”, we make all transactions take place off-chain, and then publish the results on-chain. However, we will not issue a certificate, so we need to check all transactions and see if they find inconsistencies or fraud, if this is the case, they will submit a fraud certificate. For optimistic Rollups, this model has several disadvantages.
First, there is no finality, and it must be assumed that the user is monitoring the status, because it takes several hours for the user to dispute the status update. Intuitively, Optimistic Rollup can make up for this shortcoming by being cheap, because with them, you don’t need to prove it.
But this is actually not correct, because for Optimistic Rollup, you must submit all data on the chain, that is, all transactions that have been completed. When using ZK-Rollups, we can use one piece of evidence to prove that there are many thousands of transactions. Transactions (we will soon be able to prove thousands of transactions in a batch for dYdX). Therefore, if you compare the amortized weight of a transaction in a proof with the cost of sending it to the chain with on-chain data, zk-rollup is not only more secure, but also much cheaper.
For example, I think the disadvantage between StarkWare and Optimism is that optimism is easier to learn because it is compatible with EVM. When you join Optimism, you still need to modify some things, but this is much less than removing Cairo. However, with the work on StarkNet, we hope that this shortcoming will no longer exist in the near future.
David: Brendan, can you describe your building experience in Cairo?
Brendan: We didn’t actually write code for this in Cairo, because StarkWare has more understanding and expertise in how it works. We have considered writing some Cairo contracts by ourselves, but only out of expediency, from the perspective of security and contract correctness, we basically worked with them to propose a new logic system different from our L1, because we wanted Do a similar cross-market margin system and so on. So basically working with them to come up with the logic that needs to be coded.
Most of the contracts this time were written by StarkWare. Therefore, we studied the current situation of Cairo with them and gave some feedback on how to improve it. But in this example, it is the creation of StarkWare, so I can’t comment too much.
David: Ohad, what are your thoughts on StarkWare’s strategy of adding more projects to Cairo/Layer 2? StarkWare is developing a stable Cairo compiler. When will it be available?
Ohad (StarkWare): Currently, we are adding the project to StarkEx. Since June last year, we have enabled DeversiFi, dYdX and Immutable, and hope to release several more in the rest of this year. We try to make it easy to join StarkNet in a variety of ways. Some of them may improve Cairo. A year ago, writing in Cairo was like assembly language, but now it’s more like writing C, maybe even simpler.
In addition, we also plan to compile from solidity to Cairo. We think this is just one of the few aspects that make working with StarkNet easy, so this has yet to be determined.
User: I noticed that there are index prices, oracle prices and mid-market prices. Although there are documents explaining the difference between the three, I am still not entirely sure why an index price independent of the oracle price is needed. Do you have any documents or public information about the information obtained by the two price platforms from different exchanges or the sources of oracles?
Brendan: I’m not sure whether the content on the help page is accurate, but in essence, we have discussed how the price of the oracle performs better on Layer 2, but the price of the oracle is not accurate to every minute. Therefore, the index price is mainly hosted internally by dYdX, which has two purposes different from the oracle quotation, mainly because it is more efficient than the oracle quotation.
The first is the function of triggering orders. So these are things like stop loss or take profit orders. Therefore, if you usually have trailing stop or stop limit orders, then this will trigger the index price instead of the oracle price. As I mentioned, it’s just because index prices perform much more efficiently. dYdX basically pulls prices from the exchange. Take Bitcoin as an example, refer to six to seven spot exchanges, such as Binances and coinbase.
Another use of it is to determine the interest rate of funds. The interest rate basically depends on whether the transaction price of the asset is higher or lower than the index price. So index prices need to be very precise because you are comparing fast-moving orders and fast-moving spot prices on these exchanges. So things like index prices can be updated every second and every two seconds, but the price of the oracle is usually not updated unless there is real volatility in the market.
If the price changes more than 25 points, you may see the price of the oracle updated. But the index price update frequency is much higher than this, so it provides a better experience for stop loss limit orders, and also provides a more accurate capital interest rate. So, hope this is helpful, I think we plan to build a help section and make all this information more accessible, as well as the exact exchange used by each trading pair.
User: Liquidation depends on the price of the oracle, not the index price?
Brendan: Yes.
User: Why not just rely on index prices? Is it to some extent to prevent manipulation?
Brendan: The price of the oracle is safer. The index price is basically only maintained by dYdX itself. So basically, this is a service that can track this price. If we only conduct liquidation based on this, I think this is a trade-off. Therefore, the index price is not included in the certificate or in any other content. This is an off-chain activity that we track in this way to provide a better experience.
For example, we can do everything at the price of an oracle. The reason that liquidation is not on the index price is that liquidation needs to follow a specific set of rules stipulated in the smart contract. Even if their profit margins are not high, we cannot liquidate them. All of this needs to be included in the rollup certificate itself.
User: Regarding the two products provided by dYdX, I would like to raise more academic questions about the decision made by ordinary traders between the use of margin and the use of perpetual contracts. I am sure you have considered this. I know that perpetual contracts are very popular in the cryptocurrency market. This is one of the reasons why you have actively turned to pay attention to perpetual contracts. This makes sense. One thing I want to know is what impact the perpetual market has on the spot market and vice versa. I imagine that this has been researched, maybe it’s not a big deal for ordinary traders, they just want to make money. But I want to know that I imagine the market makers you are working with are in a market neutral position, for example, if they are short while they are buying spot on the other side. What type of research have you done on the spot market, research or analysis of the difference between margin and sustainability, and how do they affect each other?
Brendan: I think your theory may be correct. You can ask David to answer in a while, but from my perspective, this is an open question. I don’t know if we have any very reliable ideas about macro trends. Personally, I may agree with some views that if the market is over-leveraged, then the price collapse will be even greater. This is the case for the market.
David: This is a big question, and I have some research on it.我的观点是，如果你回顾门头沟时代加密货币交易所的历史，你会发现中心化交易所的崛起。然后，您真正看到随着BitMEX 和Bitfinex 的兴起，中心化交易所的杠杆和保证金交易的兴起。在过去的两年中，你可以看到在像Uniswap 这样的amm 领导下的去中心化交易所中现货交易的兴起。然后我们的论点是去中心化衍生品交易所将真正起飞。
最终，如果您同时观察加密市场和非加密市场，衍生品市场的交易量要比现货市场大几个数量级。这在中心化交易所中已经非常常见了。我在本周初查看了数据，中心化交易所每天的衍生品交易量超过3000 亿，这是现货交易量的几倍。
很多价格发现实际上都发生在衍生品市场上。这通常发生在现货市场之前，因为现货市场是大量散户和机构交易的场所。所以我认为有些人会关注衍生品市场以及衍生品价格对现货的影响。我认为另一件需要注意的事情是在衍生品领域中使用的杠杆量。价格的大幅上涨或暴跌通常会导致交易员被清算或不得不回补他们的空头。
所以我认为，观察衍生品领域的杠杆量通常是市场后期走势的先行指标。然后我想回答你关于做市商的问题，我们有很多更成熟的散户用户当然还有很多做市商将衍生品视为一种获得加密货币市场中立地位的方式。因此，一种流行的交易策略是在现货市场上做多，然后在期货或永续盘上做空。它们的基础价格是上涨还是下跌，都不会影响您的整体头寸。
最终，您将获得资金费率，根据市场的不同，年化收益率在40% 到80% 之间，如果你知道如何恰当地管理抵押品，这可以被认为是无风险的。所以我们看到很多大玩家选择了热门赌局的另一边并从中获得报酬。交易员会观察不同交易所的资金利率以及利率如何随市场动向而动态变化。许多人将其视为现货市场价格走势的先行指标。
User: 您提到了衍生产品交易所，我正在观察Dune Analytics 完成的所有分析，结果显示，大约3.5％的ETH 钱包实际上与某种DeFi 协议进行了交互。 因此，实际上有少量个人真正参与过Defi。所以，现在，我看到dYdX 的数字表现不错，吸引了新用户来尝试该平台本身，但是同时也存在竞争对手。那么dYdX 是如何寻求适应的呢？还是将专注于自己的事物并观察其发展情况？ 有什么远期计划？
Brendan:我可以对这个问题泛泛而谈，但是有什么特别需要我帮助回答或者可以谈论的东西吗？
User: 随着新的衍生品交易所的推出，用户将有更多的选择，以你需要能够留住注册dYdX 的新客户。那么，有没有考虑过竞争对手推出自己的平台，以及如何留住用户？因为用户总数量是固定的，用户手里的筹码是有限的。那么，dYdX 如何才能在竞争中保持领先地位呢？
Brendan:这是个好问题。我不知道这是否特别适用于一般的加密市场。我想你是对的。在某种程度上，你是在与其他交易所争夺固定的用户池。但在很多方面，Defi 和crypto 的用户数量一直在增长，特别是在牛市时期。但我可能会给出和其他公司一样的答案：想要获得用户，留住用户，找到适合市场的产品，首先是找到市场定位，其次是超越竞争对手。
我知道这有点像一种逃避的回答，但我认为这是一个真正高质量，可靠的产品，并提供我们以前没有见过的功能。所以我们有机会成为，世界上最好的非托管交易，然后努力去中心化，我认为这将是一个更大的范式转变。但与此同时，人们使用dYdX 的原因是它是一个很好用的交易所。
我们希望成为用户期待的高质量的交易所。在过去的几年里，我们一直在想，「你们真的很适合做DeFi 交易」，但是去中心化交易所一直不被用户视为主流。所以我认为，未来几年，我们将基本上必须提高到与既定交易所一样好的水平，即使是与Binance，Coinbase，FTX 这些大型中心化交易所相比较。
我认为，总的来说，我们必须把他们当竞争对手。正如你所提到的，它现在绝对是一个热门领域，但我不知道我能说的比我们有一个更好的产品，更可靠，人们更喜欢使用比其他产品。
User: Uniswap 和Sushiswap 在代码分叉方面发生了什么？我的理解是dYdX 是开源的，对吗？
Brendan:我们的很多东西都是开源的，也有很多不是开源的。例如，我们只是在后端作为服务运行匹配引擎，这不是开源的。像智能合约之类的东西都是开源的。所以我认为在很多方面，我们确实运行了很多服务。我认为在某些方面可以给我们带来比Uniswap 更多的竞争优势，不像在Uniswap 中，只能创建合同，将其开源然后再发布。
Brendan:我认为这使得Uniswap 更容易被复制。然而，我认为我们正在运行更多的专有后端系统和高吞吐量系统，而不是仅仅依赖于人们读取和写入Layer1 区块链的状态。所以，我认为这比单纯的开源软件更有价值。这就是我对Uniswap 分叉的看法。
David:我想感谢Brendan 和Ohad 回答了所有的问题，希望这对你们有所帮助。我们将在一篇文章中回顾所有讨论过的内容，希望将来能够再有很多场这么高质量的活动。如果有其他问题，请在我们的Discord 上写下。团队也随时可以在那里回答问题。 thank you all.
