MEV Activities Since the Merge - Strategies, Searchers and Builders during Market Panic - Elaine Hu

So, I'll get to that a couple of minutes early. Elaine will then talk about MEV's post- merger activity, search engine and builder strategies during market pullbacks. Okay, great. Just checking if the sound is working. So, in the next 15–20 minutes, I'll talk about some of MEV's post-merger activities.

I'll touch on a little bit of everything, including search engine strategy, offerings, the builders' competitive market, and in particular, recent stressful or crisis events in the market. So, let's get started. Before moving on to the data, I would like to highlight the concept of MEV, as it is a theoretical value that is sometimes difficult to estimate and obtain all the data on the web. That's why we introduced the concept of Realized Excretory Value (REV). REV is the actual value that we can observe on the network thanks to the capabilities of MEV.

Before the merge, we did something similar, collecting REV from supply balance changes or miner balance changes. And after the merger, we collect the same data collect the same data . Here we use the ETH sent to the proposer as an approximation of the total REV. But keep in mind that the REV portion, which is only a slice of the theoretical MEV pie, also contains various components. For example, we only consider atomic arbitrage and MEV capabilities.

We do not consider any types of arbitration outside of a single block. There are also REVs in other domains, cross-domain, and on centralized exchanges, for which we have little data yet. So the numbers we see in the ecosystem are focused mostly on atomic arbitrage or MEV on-chain or MEV on-chain capabilities. Some of you may know about one of Flashbots' products called MEV Explorer, and the numbers from it are often quoted, but I want to emphasize: MEV-Explore only reflects a small part of the big picture. So when looking at the data from MEV-Explore, remember that it doesn't tell the whole picture.

Good. Let's take a look at MEV activity after the merger at the Layer 1 level. In 2021, we already provided similar figures using MEV-Explore. So here we show on the right the 12- show on the right the 12- month coverage of MEV activity from liquidations and arbitrage, totaling 475 million. Keep in mind that the dollar value was based on the ETH price at the time, in late 2021, which is probably much higher than it is today.

On the left is what I'll show you in the next few slides. It's been 6 months since the merge, and overall we see that around 146,000 ETH has been sent to validators. This is equivalent to approximately $260 million today. Again , looking at this number, I want to emphasize that we are using ETH paid to providers as an approximation of total MEV. We see this number accumulating since the merger.

But you see two separate bar graphs on the right, where the weekly figures are significantly higher than the other weeks. These are two events that I will talk about later in my slides. These are two market crisis events that have occurred recently. One in November—the FTX crash, another in March, more recently—the more recently—the SVB run, as well as the brief loss of stablecoin pegs. To show all participants in the supply chain, we only talk about ETH for providers, which is just one component in the chain.

And if we make the simple assumption that parties other than the provider and the search engine do not receive MEV, which is not entirely accurate right now. For example, some block developers make a profit, but let's assume that the distribution is only between search engines and providers: 48% of the MEV goes to the search engine, and 52% to the provider. So the numbers I showed on the previous slides roughly reflect that 52% on average. So what happens to the relay after the merger? We see that competition between relays is increasing.

There are more relays in the ecosystem now than there were immediately after the merger. In the beginning, we only had Flashbots, and no other relays in the ecosystem. We have a lot of relays now, and you can see that the share of Flashbots has been decreasing since the beginning of this year. Here below you also see the ETH paid out through the Flashbots relayer compared to others. You also see that there used to be a gap where the Flashbots relayer paid more than the others, but now that gap is rapidly closing.

If you look at the builder space, in the beginning, since the merger, we were dominated by Flashbots and builder 0x69, but the share of builder Flashbots started to decrease since the beginning of this year. Right now, we've only featured the top 10 builders in this space that we were able to identify. At the same time, there are many other smaller builders that I have not included, and which I will mention in the following slides. Hence the delay. What do we mean by delay?

Typically, among all the parties involved in the supply chain, we look at when the block submission occurs. This slide shows the time when the validator requests a get header from the builder. This is approximately 400 milliseconds after the slot starts. So, after these 400 milliseconds, the validator starts requesting a stake from the block builder. When the average block was ready, we see that the average winning blocks, i.

e. those that actually made it into the chain, arrive 149 milliseconds before the start of the slot. I'll just show you the winning blocks, as well as the proposed blocks that made it into the chain, compared to all the blocks submitted to the slot. This diagram actually shows a very interesting picture. Most blocks are actually served in the last 2 seconds of the slot.

And that's what we're talking about in terms of latency. So, there is a delay on the part of each party involved to submit the block as late as possible, towards the end of the slot, to include as many transactions as possible and increase the overall profit. Here, this picture shows that most of the blocks that hit the chain arrive 2 seconds before the end, and some of them arrive too late, 2 seconds after the end of the slot. But at the beginning, say the first 10 seconds, not many blocks are served. So high latency in block submission can cause serious problems.

For example, a delay in certification, as well as network overload in the last few seconds. I believe there will be speakers who will talk specifically about the issue of delay in the following sections. So I won't go into too much detail. So now let's look at two recent panic events in the market. One happened on November 9th when FTX and Three Arrows Capital crashed, and another, probably still fresh in your memory, happened just a few weeks ago when we saw the SVB bank raid and stablecoins like USDC and DAI briefly lost their peg.

Here is a chart showing the price of Bitcoin and ETH during these two crises. Interestingly, the first crisis lasted much longer, almost until early February, while the latter event recovered very quickly. So, why do we want to look at these two events and what is particularly interesting about them? Here we see that the daily MEV paid to proposers reached an all-time high since the “ high since the “ merger” during a recent event. And this is nine times more than the average daily MEV.

This happened due to the USDC decoupling event that occurred immediately after the SVB bank raid . Whereas previously, during the FTX crash, we saw slightly less severe MEV rates for proposers, but still six times higher than average. Also, considering MEV from the perspective of searchers, we collected data from EigenPhi. This shows arbitrage, sandwich attacks, and liquidations during these two events. You may also notice that this chart shows the same pattern, where arbitrage volume skyrocketed that day.

Interestingly, cross-chain cross-chain data analysis shows a slightly different picture. You can see that the FTX event had a larger cross-chain impact, while the recent SVB bank raid and stablecoin decoupling were more localized within the Ethereum network. This is interesting data from Sam Ottix regarding MEV crosschain. In particular, we will look at how much ETH was paid to proposers on these two days, what the competition among builders looked like on these days, as well as the delay in submitting blocks. And this applies to Ethereum.

We will also look at cross-chain strategies over these two days. So, on March 11th, when the bank was raided, we see that 6,000 ETH was paid to the proposers paid to the proposers . This is equivalent to or even greater than the average weekly total of ETH paid to proposers. Below you can see that the weekly average is around 4,000, while in just one day, March 11th, 6,000 were paid out. Similarly, on November 9th, when FTX crashed, around 4,000 ETH were paid out .

So, this is an extremely large amount of MEV activity in one day that I'm highlighting here. Um, on the right I have a breakdown by builders by total ETH paid to proposers. So, in November, we saw the leading builders—Flashbots, Beaver Build, and Builder 69— Beaver Build, and Builder 69— dominate the market, taking over 60% taking over 60% share. However, during the recent event, the share of top builders has decreased and we see more players entering the market, with smaller builders taking up 60% of the total ETH paid to proposers. So, we see some decentralization or diversification in the builders market during the latest event compared to November.

What about search engine strategies search engine strategies ? We will also look at these two days. On the left you can see the arbitrage for these two days. And on the right you can see the liquidations. Note, I already mentioned that MEV-Inspect, on which the MEV Explore dashboard is based, only takes into account liquidations and arbitration.

So this is just a small part of what happens in a day. It is worth noting that EigenPhi also provides this data, covering a wider range of MEV types. They also take sandwich attacks into account, and probably include more protocols. But here I am only selectively looking at the data from these two days with MEV Explore. You can see that most of the arbitrage transactions took place between Ethereum, Bitcoin, and stablecoins Bitcoin, and stablecoins .

From the perspective of the tokens . From the perspective of the tokens participating in the arbitrage, the situation in both events is very similar. However, when it comes to liquidations, they depend more on the specific event. As you can see, during the recent event, the Curve pool received a much larger share of the total volume, while the November event was all about arbitrage between different tokens. Good.

Regarding searcher strategies by protocol, here we also look at MEV Explore data, which includes nine different protocols. Comparing these two days , we see essentially the same picture. In November, when FTX crashed, most of the arbitrage was on Uniswap V2 and V3, while during the latest event we see an increase in the percentage of arbitrage in the Curve stablecoin pool. As for liquidation, the latest event did not cause a large liquidation in dollar terms. This is quite obvious as the latter event does not have a strong correlation with liquidations, while the FTX crash had a broader impact on the market, where more liquidations occurred.

OK. Regarding the delay, on the left we see the November FTX event, and on the right we see the and on the right we see the recent event. Interestingly, the graph on the right is very similar to the general graph I showed at the beginning of this presentation. This likely indicates that the increase in latency is more noticeable compared to the beginning of the merge period, as most blocks in November were served fairly evenly over 12 seconds, whereas now most blocks are served in the last 2 seconds. So, in just a short span of 4 months, we see that the delay has become more serious.

So this is probably not due to a market panic, but rather a time factor as relays and builders become more sophisticated and compete for block addition speed. Now, looking at cross-chain looking at cross-chain interactions, we see two different types of events occurring in this area as well. The most popular arbitrage tokens are very similar to those used on the same network, such as Ethereum. It can be seen that most transaction paths were initiated from stablecoins, Bitcoin or ETH. It is also very similar in terms of percentage.

So, the situation probably doesn't vary much depending on market events. I want to emphasize again that cross-chain MEV in dollar terms was significantly higher during the FTX crash, while the recent event has a significantly lower MEV. Looking at the networks and specific paths, we see that Ethereum and BSC are the networks most involved in arbitrage. It is also worth noting that in November we observed activity on the Polygon and Arbitrum networks. Whereas during the recent event, arbitrage occurred mainly between Ethereum and BSC.

OK. Finally, a little more detail on the top 10 arbitrageurs by profit; We have shown the top 10 arbitration agreements held on various networks. So, this is exactly the way that bots have modeled themselves to detect any arbitrage opportunities. It is worth remembering that this data is not real data on interchain MEV. They are based on simulations of eight different networks, about 40 decentralized exchanges, and 60 tokens.

So there may be inaccuracies in the simulation , and due to missed arbitrage opportunities or data, the figures may be even higher. In summary, I have talked a lot about the various actions and trends of MEV in L1 and cross- trends of MEV in L1 and cross- domain networks after the merger. It can be concluded that estimating MEV is a very difficult task. We only have limited Ethereum data and network data. There is a lack of sales data for any MEV assessment within the scope of centralized exchange operations.

We also see that after the merger, more repeaters and builders have come into the ecosystem, and the space has become more diversified. Interestingly, the market share of leading builders during periods of high MEV actually decreased. This also suggests that the builder market has become more diversified since the merger. Regarding arbitrageur strategies, we do not see much difference between different market crises. Most arbitrage strategies focus on transactions between Ethereum, Bitcoin, and stablecoins Bitcoin, and stablecoins .

Elimination strategies are actually based on specific events. So if a crisis involves liquidating certain assets, you will see the MEV for liquidation increase. But in the case of recent events, this is not the case. Cross- this is not the case. Cross- domain strategies have the same tendency.

MEV activity increased dramatically during moments of crisis. But in terms of scale, it is on average much higher than the daily one. Probably because there are fewer daily cross-domain arbitrage transactions than in Ethereum, but during big events, bots look for opportunities everywhere. This is why there is a much larger spike in activity during such events. Furthermore, strategies differ depending on market crises.

They depend on specific events. Finally, I would like to note that due to the lack of data, we have a lot of room to improve the assessment of cross- assessment of cross- domain MEV and MEV between centralized and decentralized exchanges, so there is still a lot of work ahead to collect accurate data on the activity and total value of MEV. So, here is a list of links, as well as some interesting reading material that I recommend. Finally, special thanks to Tina from Flashbots, as well as Igal from Blocksroute, Metrika, Semiotic, and Blocknative for their help in preparing slides, charts, and data. Mhm.