Coin Metrics' State of the Network: Issue 114

Tuesday, August 3, 2021

Get the best data-driven crypto insights and analysis every week:

Analyzing Tether and USDC Usage Patterns

By Nate Maddrey and Kyle Waters

The total supply of stablecoins is near 110B today, almost a 4x increase over the supply at the beginning of 2021. But this growth has not been distributed equally across stablecoins. Amidst an uncertain crypto market, the stablecoin market is also shifting. 

The supply of USD Coin (USDC) has grown remarkably in 2021 from ~4B at the beginning of the year to over 26B today. USDC’s growth accelerated following the May crypto crash - USDC’s total supply has increased by about 75% since May 11th. Tether (USDT) still remains the dominant player in the stablecoin market with a total supply of over 64B, but USDT growth has flatlined since May. USDC is the second largest stablecoin behind Tether, and its share of total stablecoin supply is growing. 

Source: Coin Metrics Network Data Charts

While Tether’s total supply growth has stalled, its free float supply has decreased over the last month. “Free float” supply is a metric developed by Coin Metrics that excludes supply that is considered illiquid. Tether’s decrease in free float supply represents supply that has been redeemed and sent back to Tether’s reserves. This free float decrease has been especially noticeable on Ethereum - the free float supply of Tether issued on Ethereum (USDT_ETH) has decreased by about 1B over the last 30 days, while the total supply has remained flat.

Source: Coin Metrics Network Data Charts

USDT_ETH’s daily usage patterns have also shifted recently. Historically, the bulk of Tether activity has mostly occurred during Asian business hours. The heatmap below shows the hours when Tether issued on Ethereum (USDT_ETH) is most frequently used. Darker reds correspond to the hours (in UTC) with a higher percentage of daily activity. 

Throughout 2020 most activity was concentrated between 2:00 and 14:00 UTC, matching Asian / European business hours. The period from 6:00-8:00 UTC was especially busy, as indicated by the darkest red regions. For context, continuous trading at the Hong Kong Stock Exchange occurs from 1:30 UTC to 8:00 UTC, and the London Stock Exchange is open from 8:00 to 16:30 UTC. But over the course of 2021 this distribution has shifted slightly later in the day towards European/US market hours. 

Source: Coin Metrics Network Data Pro

There are several potential contributing factors. While there’s over $30B worth of Tether issued on Ethereum, Tether is also issued on many other blockchains including Tron and Solana. Tether is typically used extensively for trading and some of that trading activity may be shifting to other blockchains that can offer lower fees than Ethereum. Additionally, the miner and investor migration out of China due to new government regulations may be causing a drop in Asia-based Tether activity. However, this would only explain changes since May and not earlier. USDT_ETH is also increasingly being used as collateral in DeFi protocols which may contribute to some of the usage shifts. 

In contrast, USDC activity tends to follow US market hours, though it is slightly more distributed. The majority of daily activity tends to be between 14:00 and 22:00 UTC (for context, US equity markets open at 14:30 UTC and close at 21:00 UTC). 

Source: Coin Metrics Network Data Pro

This distribution has also been fairly constant since 2020, suggesting that Tether is still the most dominant stablecoin used in Asian markets, although the overall activity may have decreased. In comparison, Bitcoin and Ethereum use tends to be far more equally distributed by time of day.

USDC and other stablecoins are being adopted for many reasons, but one major benefit of stablecoins over traditional settlement infrastructure (e.g. Fedwire) is that they run on crypto-native payment rails that operate 24/7, 365. The heatmap below breaks down USDC activity by day of week. On-chain data show that USDC is mainly used during weekdays, but has some activity during the weekends when traditional infrastructure is largely unavailable.

Source: Coin Metrics Network Data Pro

As stablecoins become a larger part of the total crypto market they’re also gaining more attention from the outside world. Regulators are turning their attention towards stablecoins at an increasing rate, and the impending rise of CBDCs may also impact stablecoins. It will be crucial to continue to study stablecoin usage patterns moving forward as stablecoins become an even larger part of the crypto ecosystem.

To follow the data used in this piece and explore our other on-chain metrics check out our free charting tool, formula builder, correlation tool, and mobile apps

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

Bitcoin (BTC) and Ethereum (ETH) both had strong weeks with BTC surpassing the $40K mark and ETH clearing $2.5K. Hash rate continued to recover as mining operations start to get back online after migrating out of China. Bitcoin hash rate grew 9.7% week-over-week while Ethereum hash rate grew by 3.8%. On-chain activity has also risen with active addresses growing ~8% and ~9% over the last 7 days for BTC and ETH, respectively. 

Network Highlights

EIP-1559, a highly anticipated and deliberated Ethereum Improvement Proposal (EIP), is set to go live this week (at block 12,965,000) as part of the series of London Ethereum upgrades. The current Ethereum transaction fee mechanism has sometimes been a source of confusion and frustration for users, especially after ETH fees hit new all-time highs earlier this year. EIP-1559 brings a fundamental change to how Ethereum fees work. 

Source: Coin Metrics Network Data Charts

Breaking Down the Core Features of EIP-1559

The original EIP proposed in April 2019 is summarized as (emphasis added) “A transaction pricing mechanism that includes fixed-per-block network fee that is burned and dynamically expands/contracts block sizes to deal with transient congestion.”

Below we break down each part of this summary to illustrate how EIP-1559 will work. 

“Transaction pricing mechanism” 

To send a transaction or interact with Ethereum decentralized applications (dapps) users need to pay a fee. Ethereum fees are commonly referred to as “gas.” Similar to how a car needs gas to run, Ethereum applications need gas in order to be executed. 

Currently, Ethereum employs what is known as a first-price auction to determine gas prices which can create uncertainty and inefficiency. Imagine being at a busy airport trying to hail a taxi. But instead of waiting in line to get a taxi you blindly bid the highest you would be willing to pay to complete that ride, without knowing what others in line were bidding. The taxi drivers, on the other hand, could see all the incoming bids and maximize their profits by selecting the top bidders. 

This is a simplification of Ethereum’s current fee mechanism but ultimately users must think strategically about what other people will bid, which often results in overbidding or underbidding and leads to high fee volatility. For example, the chart below shows the mean and median transaction fee paid in GWEI by block for a sample of ~5,000 blocks on July 25, 2021. Note the large outlier blocks in which the mean (red) is far higher than the median (green). This suggests that some transaction senders were overpaying, since the median fee would be sufficient to get a transaction included in the block.   

Source: Coin Metrics Network Data Pro

“Fixed-per-block network fee that is burned” 

To improve fee predictability to the Ethereum user experience EIP-1559 introduces a base fee at each block. The base fee is a required payment to be included in a block and is programmatically determined based on the previous block. This in effect automates the gas price bidding system. Under the chosen parameters, the base fee cannot move up or down by more than 12.5% from one block to the next. 

This provides something that is more akin to a predetermined list price that a user can reject or accept. However, the user has the option to also add a tip. 

Potentially the most discussed component of EIP-1559, the base fee will be burned rather than paid out to miners. This is promising for Ethereum’s supply economics as it will permanently remove some ETH and lower supply inflation. Using some basic assumptions in which 75% of fees are burned, the chart below presents a historical scenario of ETH supply issuance with EIP-1559. Note that in times of high network congestion daily issuance can turn negative. 

Source: Coin Metrics Network Data Charts

“Dynamically expands/contracts block sizes” 

To determine how the base fee changes from block to block, the protocol needs to have an estimate of demand for block space. EIP-1559 completes this by introducing a target block size. In short, the maximum Ethereum block size will double from its current limit but target 50% capacity. If the preceding block was larger than the target block size (i.e. more than 50% full), the base fee increases and keeps increasing until block size falls back to its target. This escalating base fee eventually makes it too expensive for some users to transact, reducing congestion and causing block fullness to naturally move back towards 50%.

A common misconception of EIP-1559 is that it is intended to address high transaction fees and bring down average fees paid by users on-chain. But high fees are ultimately issues of scalability and are not functions of an inefficient or unpredictable fee mechanism. Scalability is being addressed via layer 2 (L2) solutions and the eventual upgrades planned for Ethereum 2.0. However, fee volatility should decrease on a block-by-block basis as a result of better fee predictability. For a more detailed analysis of Ethereum’s gas fee mechanism see The Ethereum Gas Report.

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Check out our market-data focused newsletter State of the Market, featuring weekly updates on market conditions.

As always, if you have any feedback or requests please let us know here.

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 113

Tuesday, July 27, 2021

Get the best data-driven crypto insights and analysis every week:

Understanding Total Value Locked (TVL)

By Lucas Nuzzi, Antoine Le Calvez, and Kyle Waters

Decentralized Finance (DeFi) has become one of the industry’s most vibrant topics, with dozens of new projects launching on a monthly basis. DeFi applications essentially enable the creation of financial contracts that are automatically executed. Broadly speaking, these contracts facilitate the issuance, lending, trading, and management of cryptoassets.

Given the broad scope of DeFi applications, it is difficult to measure the adoption of the DeFi theme as a whole. After all, trading and lending involve different operations that are not fully comparable. In order to address this, the industry has converged on a metric called “Total Value Locked”, or TVL, to measure the adoption of DeFi projects. 

Most DeFi applications, whether they facilitate lending or trading, require a deposit of collateral in the form of a cryptoasset, like a stablecoin. The “Total Value Locked” of a protocol is simply taken as the sum dollar valuation of all collateral deposited in that specific application, regardless of its functionality. As such, TVL enables money markets like Aave to be compared with Decentralized Exchanges like Uniswap.

Since 2019 DeFi has experienced exponential growth. TVL has become the de-facto way to measure DeFi adoption, and one of the most requested metrics at Coin Metrics. In this post, we would like to share some of the challenges that prevent the accurate calculation of TVL, as well as the critical shortcomings when using this metric to evaluate DeFi protocols.

We identified three major challenges that make computing a robust TVL metric a difficult task.

Protocol variety: the myth of “Total”

Decentralized finance is still a very young field with protocols and applications launching and disappearing every day. Some of these launches are mere copies of already existing protocols or new versions within an existing system, while others are entirely new creations. Each of them complicates holistic estimates for a given smart contract blockchain.

Protocols can become overnight successes, rallying billions in collateral in a matter of days. For example, Sushiswap, a clone of Uniswap, rose to near-instant fame in September 2020, as its TVL went from a few thousand dollars to over a billion overnight:

What explains this growth rate? In essence, incentive structures in DeFi protocols are entirely malleable. SushiSwap came out of nowhere and attracted over a billion dollars in deposits via an aggressive issuance schedule that favors early adopters with newly minted SUSHI tokens.

This approach set a precedent that will likely be repeated ad infinitum. Given the frequency of protocol clone launches, it has become nearly impossible to perfectly track all collateral allocated to a blockchain in real-time. Observers like Coin Metrics are required to pick which protocols to individually track TVL for since each requires some amount of manual work to be integrated.

The frequency of new protocol launches leads to a natural underestimation of the total value used as collateral across all DeFi applications by all data providers. In order to accurately calculate TVL for a platform like Ethereum, providers must constantly re-evaluate past measurements to reflect the addition of new protocols and collateral types. With DeFi now being aggressively pursued in new smart contract platforms, this fast rate of new launches undermines any approach to have correct protocol-wide TVL estimates.  

Beyond the issue of new protocol launches, another complicating factor is that existing protocols can also mutate. In order to account for such mutations, new versions and contract deployments must also be constantly monitored. Uniswap, for example, is already at its 3rd iteration, with collateral being tracked slightly differently for each version. Accordingly, TVL for Uniswap is the sum of collateral allocated to each of its versions, which must be individually assessed.

It is possible that DeFi stabilizes around a set of norms or standards in the future. If such standardization were to happen, that could make the task of integrating new protocols easier. However, standardization is no panacea since there is no guarantee that all protocols would follow standards accurately. As we have seen with the proliferation of the ERC20 standard, many variations exist that still require manual review. As such, it is unlikely that the normalization of DeFi will grant data providers a leap of performance when it comes to protocol additions in the short to medium term. 

Collateral type variety: the myth of “Value”

Decentralized finance protocols support a near-infinite variety of assets that can be used as collateral. While some protocols restrict the collateral types, many do not.

(includes data from Uniswap v1/v2/v3, Sushiswap, Curve, Aave v2, Compound, Maker)

This chart gives a lower bound on the real number of distinct assets used as collateral in DeFi applications. It is not a holistic estimate since it does not include data from all DeFi protocols and is limited to ERC-20 tokens only. Nevertheless, this lower bound provides a glimpse at the impact of tokenization and the rapid growth of collateral types that can now be used in DeFi.

The sheer scale of collateral types complicates the value estimations. All of these assets can be traded across multiple venues ranging from centralized, off-chain, exchanges to decentralized, on-chain, protocols. The collection of pricing data from all of these venues is a herculean task that, just like protocol integration, does not scale well. At the same time, this is a requirement so that the asset used as collateral can be priced correctly via an index value that accounts for each venue.

Even if a data provider were to have the bandwidth to produce index values from all possible trading venues, it is hard to take all data collected at face value. Much like the problem of correctly calculating a cryptoasset’s market cap, pricing data in DeFi liquidity pools can be manipulated and ultimately undermine value measurements.

Robust price sources like Coin Metrics reference rates cover at most the top few hundreds of assets. For the remaining long tail of assets, it is possible to use on-chain exchanges to estimate their current price, but there’s no guarantee that they are traded with enough frequency to give accurate prices, or that the liquidity in these protocols was added organically.

Rehypothecation: the myth of “Locked” 

Finally, the most subtle yet important challenge with using TVL comes with understanding the makeup of assets used as collateral. When evaluating a protocol’s TVL, one might assume each unit of value deposited as collateral is being exclusively used within the protocol. In other words, the assets are “locked” in that they are strictly being used in the context of the application and nowhere else.

However, due to how DeFi money markets are designed, this assumption is wrong. DeFi enables the creation of asset derivatives that rehypothecate collateral. Put simply, the collateral used in one application can be used in another, which can then be used in yet another, and so on and so forth. There are DeFi applications specifically designed to enable rehypothecation so that their users can get leverage. While the existence of this is nothing new, it should nonetheless challenge one’s understanding of what “locked” collateral represents.

In brief, some assets used as collateral in DeFI applications are derivatives that represent existing claims on other collateral. This results in a multiplier factor that can drastically increase TVL estimates since both real and rehypothecated collateral are being counted. Existing approaches to the calculation of TVL fail to discern between the two. As such, collateral can be considerably overcounted depending on the protocol. 

In order to illustrate this, take a look at the following example:

  • A user deposits $1,500 worth of Wrapped Ether (WETH) into Maker to get a loan in the form of $1,000 worth of DAI (150% collateralization ratio).

  • The user then deposits this newly minted DAI, as well as another $1,000 worth of USDC in the Uniswap V2 USDC-DAI pool. In return, the user gets Liquidity Provider (LP) tokens representing that $2,000 stake of that pool’s liquidity.

  • The user can then redeposit these LP tokens into Maker to get another loan of $1,960 of DAI (102% collateralization ratio)

From a naive perspective, TVL could be computed as:

Yet, a more sophisticated approach would only count the $1,500 of Wrapped Ether and $1,000 of USDC as the “real” collateral giving a TVL of $2,500. This approach would not include assets that are claims to other collateral such as DAI (which is minted as a loan against collateral), and Uniswap DAI/USDC LP tokens (which represent a claim to the liquidity held by the Uniswap V2 DAI/USDC pair).

This adds additional complexity as determining whether or not an asset used as collateral adds “hidden” leverage to TVL values. 

The Search for Better DeFi Metrics

In order to better reason about DeFi systems and measure them appropriately, it can be helpful to think of DeFi assets as neo asset-backed securities (ABS). These are a type of financial derivative that represent a claim to a pool of collateralized assets. In the context of DeFi, these derivatives provide the foundation to the trading, lending and management of cryptoassets. Relative to the legacy systems used to issue ABS, DeFi attempts to bring higher transparency and automated risk management.

In this vein, what Total Value Locked is measuring is the total size of a levered market. As covered in this report, that figure can be misleading as it is inflated by a leverage multiplier, carries high price sensitivity, and is far from holistic. Without knowing what that multiplier is, it becomes impossible to measure the health of the system. Most importantly, the lack of this data prevents a systemic analysis of a system’s sensitivity to price shocks, which is critical intelligence for these types of systems.

For these reasons, it is of the utmost importance for us to distinguish between “real” and rehypothecated collateral before producing our own TVL estimates. Similarly, it is important to track these figures in the protocol’s “native units”, which removes price sensitivity and provides a better picture of an application’s growth. Beyond better approaches to TVL, another potentially interesting metric to cover is a more simple equivalent to DeFi’s “Open Interest” where, instead of value, the total number of contracts supporting an application is counted.

Needless to say, it is challenging to pursue the creation of all of these metrics at once. In order to better automate the data collection process, we are building a new set of tools to parse smart contract data in a more scalable way. Given the challenges associated with holistic measurements, our expansion of DeFi metrics will focus on risk management at the application level, as well as trades data from high profile AMMs. 

Conclusion

In conclusion, Total Value Locked is a deceptively complicated metric hiding under a benign name. Each word that make up its name brings its own set of challenges:

“Total” means tracking all the versions of a protocol and now even some versions of it on multiple chains (Ethereum, Binance Chain) as well as second layers like Matic or Fantom.

“Value” means finding a robust price for each of the thousands of assets that can be used as collateral. 

“Locked” is a misnomer as in most protocols liquidity can be added or removed quite quickly. It also means untangling the links between each asset to avoid double or triple counting.

As an industry, it is important to converge on better methodologies to contextualize and caveat growth in DeFi applications. This will be a collaborative process, so we look forward to contributing to better metrics, and learning from the community as a whole. 

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

Bitcoin (BTC) and Ethereum (ETH) both rose midweek following a high-profile conversation between Cathie Wood, Elon Musk, and Jack Dorsey at The B Word Conference (Coin Metrics’ co-founder Nic Carter and Sr. Research Analyst Nate Maddrey also participated in the conference, see CM updates below). 

Among new information, Musk hinted that Tesla may once again start accepting bitcoin as payment conditional on sustainable energy continuing to constitute a greater makeup of the energy mix of Bitcoin’s hash rate. Musk additionally disclosed he personally holds bitcoin, ether, and dogecoin, and also disclosed that SpaceX holds bitcoin (amount not revealed).  

Bitcoin also continues to naturally adapt to changes in Hash Rate. Bitcoin’s mining difficulty, which programmatically adjusts roughly every 2 weeks (2,016 blocks) to target a 10-minute average time between blocks, decreased 4.81% on July 17. This followed the near 28% fall in difficulty on July 3rd, the largest in Bitcoin’s history:

Source: Coin Metrics Network Data Pro

This also marked the 4th straight decline in difficulty, the longest such streak since 2011 when difficulty decreased 8 times in a row:

Source: Coin Metrics Network Data Pro

This is simply a sign that Bitcoin is naturally readjusting to relatively large changes in hash rate, as designed. 

Network Highlights

Maker, one of the earliest Ethereum-based decentralized finance platforms, announced last week that the Maker Protocol will complete a long-stated goal to become a fully decentralized autonomous organization (DAO). In a blog post, Maker Foundation CEO Rune Christensen formally disclosed that the Foundation, which has existed to help promote protocol growth and DAI stablecoin adoption, will dissolve “within the next few months” leaving full governance of the protocol to the DAO and the community of MKR token holders. The Maker Foundation had already taken steps in this direction in recent years, giving full control of the Maker smart contracts to holders of MKR in March 2020 and returning 84K MKR to the DAO in May 2021, but formal dissolution is the final step in this process. 

MakerDAO governance is completed by MKR holders who “lock-up” their tokens into a designated Voting Contract. Voting outcomes are determined based on the quantity of MKR tokens in support of a given side (e.g. changing important protocol parameters such as acceptable collateral types or the debt floor/ceiling). Given the importance of MKR stake in governance, the distribution of MKR supply is an important measure to follow on-chain. Historically, MKR supply has been somewhat concentrated in a few addresses. However, the number of addresses with smaller MKR balances has been growing. 

Source: Coin Metrics Network Data Charts

The number of addresses holding at least 1/10B of total MKR supply but no more than 1/100M increased from ~115K to ~175K in 2021 through July 20th.

Additionally, the percentage of MKR supply held by the top 100 addresses is still ~82% but is decreasing. From an economic standpoint though, the largest MKR holders should have the largest incentive to safeguard and grow the protocol.

Source: Coin Metrics Network Data Charts

MakerDAO is in charge of managing a growing supply of DAI in circulation. DAI supply (ERC-20) has surpassed 5B in 2021, growing steadily throughout the year. Accompanying this growth in supply, the value of on-chain DAI transferred is increasing while the price of DAI has remained close to its peg, even during times of market stress.

In the long-run, MakerDAO expects to foster three key governance elements to ensure success of the protocol: elected paid contributors (EPCs), Maker improvement proposals (MIPs), and the ability for MKR holders to delegate their votes. These refinements may help the DAO address potential challenges to protocol health such as excessive borrowing/lending of MKR or collateral concentration. 

Measures of MKR supply distribution, DAI issuance, and other on-chain analytics will become increasingly important as well for self-sufficient and effective governance of the protocol via the DAO and Maker community. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics’ Co-Founder Nic Carter and Senior Research Analyst Nate Maddrey both presented at The B Word Conference on July 21, 2021. You can view a recording of their presentations debunking common Bitcoin myths here

  • Check out our market-data focused newsletter State of the Market, featuring weekly updates on market conditions.

As always, if you have any feedback or requests please let us know here.

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 112

Tuesday, July 20th, 2021

Get the best data-driven crypto insights and analysis every week:

Crypto Google Search Trends Analysis

By Nate Maddrey and Kyle Waters

Google search data is often used to analyze a wide-variety of trends. Although a Google search doesn’t necessarily translate into a crypto purchase, analyzing search data for crypto-related keywords can help shed light on retail interest in a variety of cryptoassets. 

In this week’s State of the Network we compare Google Search Trend data to our network and market data to analyze how Google search interest relates to price and usage changes. 

Bitcoin Search Trends

Worldwide Google search volume for Bitcoin shows that interest levels closely followed BTC’s rapid price movements this year and tend to track major changes in price historically. While Google Trends data is not absolute search volume for Bitcoin, it does show the relative level of popularity for Bitcoin searches over time. 

Sources: Coin Metrics Reference Rates, Google Trends

Relative search interest this year for Bitcoin has not yet surpassed levels achieved in late 2017. This might be a sign that many retail investors and the general public were already aware of Bitcoin prior to the recent price movements this year. Institutional adoption is a big reason for Bitcoin’s recent successes in 2020/2021 which will not be captured easily from Google search interest. 

Ethereum Search Trends

While Bitcoin’s search interest did not surpass levels of popularity in 2017, Ethereum reached a new peak in worldwide Google search popularity earlier this year. Supported by mass interest in NFTs and the continued adoption of DeFi, global interest in Ethereum has been at an all time high in 2021.

Sources: Coin Metrics Reference Rates, Google Trends

With retail investors likely playing a large role in Ethereum’s success in 2021, it is unsurprising to see such high interest via Google search. Surging Google search interest for MetaMask, an ETH wallet that counted five million monthly active users in April, is also consistent with higher ETH usage. Institutional adoption of ETH is in its early stages but might accelerate like Bitcoin following this recent burst of interest. 

Dogecoin Search Trends

Dogecoin (DOGE) had very little relative Google search interest before 2021. Then in January DOGE suddenly started to skyrocket. DOGE searches surged the week of January 24th, as DOGE price climbed from under a penny to close to 5 cents. DOGE search volume peaked the week of May 9th, right before the crypto-wide market crash. 

Sources: Coin Metrics Reference Rates, Google Trends

Interestingly, Dogecoin searches initially spiked the same week that popular meme stocks GME (Gamestop) and AMC (AMC Entertainment Holdings) did. This coincided with GME’s sudden surge and Robinhood’s subsequent halting of GME and AMC trading on January 28th. Robinhood is also a popular platform for DOGE trading - according to Robinhood’s latest S-1, DOGE trading accounted for “34% of our cryptocurrency transaction-based revenue” in Q1 2021, up from 4% in Q4 2020.

Source: Google Trends

Stablecoin Search Trends

A big development in the crypto ecosystem over the last year has been the continued issuance and adoption of stablecoins. Search interest for stablecoins has surged like never before in 2021, eclipsing the fall 2018 launch of USDC and the June 2019 announcement of Facebook’s Libra (rebranded to Diem) project and subsequent update in April 2020. This search popularity is consistent with the growth in the supply of these tokens including Tether, USDC, DAI, BUSD, HUSD, and PAX. 

Sources: Coin Metrics Network Data Charts, Google Trends

This topic is likely to remain a popular one, especially in the US, with the Federal Reserve expected to release a report in early September on the potential for a digital version of the US dollar (CBDC), stablecoins, and cryptoassets in general. Ahead of this paper, Secretary of the Treasury Janet Yellen has also convened the President’s Working Group on Financial Markets (PWG) and other agencies to discuss stablecoins, a sign they are top of mind for US authorities. Finally, the expected public listing of USDC operator Circle via a SPAC merger will likely drive new conversations around stablecoin transparency. 

Geographic Insights

While global search interest for Bitcoin is relatively lower compared to earlier years, there are geographic pockets where interest in Bitcoin has never been higher. Search interest for Bitcoin in El Salvador boomed following the June announcement that bitcoin would be accepted as legal tender. 

Source: Google Trends

Note that because Trends data is a relative measure to max popularity, the small interest levels prior to this announcement do not necessarily mean Bitcoin was unpopular or unheard of in El Salvador. More likely, the spike just reflects an exceptional increase in search volume for Bitcoin in the country.

In fact, El Salvador is ranked first among all regions with the highest proportion of Google search interest in Bitcoin over the last 90 days ended July 18, 2021.

Source: Google Trends

Interestingly, the top countries are fairly constant across different time windows. Future analysis of global Bitcoin adoption would help identify the local forces driving interest in these individual countries and others. 

To follow the data used in this piece and explore our other on-chain metrics check out our free charting tool, formula builder, correlation tool, and mobile apps

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

The total supply of Tether (USDT) was again stagnant over the last week and hasn’t grown since the end of May. Because USDT is mostly used during Asian and European market hours, this might signal weaker demand in Asia, likely due to ongoing regulatory pressure.

Meanwhile, USDC supply continues to expand, growing another 2.5% in the last week. While USDT supply has been flat since June 1, USDC supply has grown by roughly 4B from ~22B to ~26B. This may point to changing dynamics in stablecoin use and supply.

Network Highlights

China’s sudden crackdown on Bitcoin mining in Q2 2021 (see our Q2 recap here) left miners with no choice but to shut down operations and move resources elsewhere. Bitcoin’s hash rate fell by ~50% as a result, but appears to be rebounding (2-week moving average). 

Source: Coin Metrics Network Data Charts

This might suggest that some Chinese miners have successfully relocated and restarted their mining rigs. In addition, Ethereum’s hash rate has also started to bounce back.

Source: Coin Metrics Network Data Charts

Interestingly, new data from the Cambridge Bitcoin Electricity Consumption Index (CBECI) show that China’s share of total Bitcoin hash rate was declining prior to the government’s mining shutdown. This indicates that China was already becoming a less popular destination for new mining operations. The data also imply that the US and Kazakhstan have been gaining most in the share of hash rate up to April 2021. It’s still unclear where Chinese miners will relocate after the recent ban, but gains in hash rate share for these two locations might be a good signal to follow. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Check out our market-data focused newsletter State of the Market, featuring weekly updates on market conditions.

  • We’re excited to announce the Coin Metrics mobile app. View real-time cryptoasset pricing and relevant on-chain data in a single app!  Download for free here: https://coinmetrics.io/mobile-app/

  • Coin Metrics is actively hiring for many positions including in engineering and data science. If you are interested in learning more about our open roles please visit our careers page.

As always, if you have any feedback or requests please let us know here.

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 111

Tuesday, July 13th, 2021

Get the best data-driven crypto insights and analysis every week:

The State of the Network Q2 Wrap-Up

By Nate Maddrey and the Coin Metrics Team

After a strong start to the year, the second-quarter of 2021 was headlined by a crypto-wide crash. But despite the crash, many assets finished the quarter in the green. In this special edition of State of the Network we take a look back at the biggest storylines of Q2, 2021.  

April Boom

It was a big quarter for Ethereum (ETH), finishing up by 13.2%. Opening April at $1,971, ETH surged to a new all-time high of $4,155 on May 11th before the markets came crashing down.

ETH benefited from a renewed surge of retail interest which was partially driven by the rapid rise of NFTs. Although NFT media interest peaked in March, it helped bring unprecedented mainstream attention to Ethereum which led to a flood of new users. 

While cryptoart trading volume has since cooled off, the broader NFT ecosystem continued to grow in Q2. NFT marketplace OpenSea recorded it’s biggest month ever (in terms of sale volume) in June thanks to a boom in NFT avatars as well as projects like Art Blocks.

Source: Google Search Trends

Thanks to the retail surge, the amount of addresses holding at least 0.1 ETH grew from 4.58M to over 5.20M in Q2. 

Source: Coin Metrics Network Data Charts

Many ETH adjacent tokens also did well during Q2. MATIC, the native token of Ethereum scaling solution Polygon, finished the quarter up 227%. Ethereum competitors like Cardano (ADA) and Solana (SOL) also finished the quarter in the green. Ethereum Classic (ETC) even got in on the act, closing the quarter up 297%.

In addition to ETH, Dogecoin (DOGE) and other meme coins peaked on May 11th. Fueled on by Tweets from Elon Musk, DOGE exploded in Q2 2021, closing the quarter up 391%. This rapid growth is also reflected on-chain. The number of addresses holding at least 1 DOGE increased from 3.09M on April 1st to over 3.7M on June 30th.

Source: Coin Metrics Network Data Charts

Collectively, smaller-cap assets reached new peaks in May. As a result, Bitcoin dominance fell to its lowest levels since July 2018.

Source: Coin Metrics Network Data Charts

May Crash

Then, on May 12th, things suddenly changed. 

First, Tesla announced that they would no longer be accepting BTC as payment. While this news initially spooked the market, even more devastating news soon followed as reports started to surface that China was cracking down on Bitcoin mining and trading. Although China has a long history of crypto regulations these new regulations appear more severe than previous iterations. On May 19th, after price unexpectedly dropped below $40K, BTC experienced a liquidation cascade that caused price to plummet close to $30K. 

Source: Coin Metrics Reference Rates

Propped up by record high levels of leverage the markets were in a precarious position before the crash. BTC open interest on Binance, BitMEX, Kraken, Deribit, FTX, Huobi, OKex reached over $4B in late April, indicating that there was likely high amounts of leverage in the system.  Further complicating things, a big difference between this cycle and the 2017 run was the ability to use leverage on projects outside of the top 10. Thanks to burgeoning derivatives infrastructure some smaller-cap assets saw relatively high amounts of open interest which likely contributed to the sudden crash. 

Source: Coin Metrics Market Data

Miner Migration

China’s crackdown on miners has started an unprecedented Bitcoin hash rate migration. Faced with sudden bans Chinese miners are in the process of relocating and setting up their operations abroad in more crypto friendly territories. Although it remains to be seen exactly where they end up settling, there’s reportedly been a large influx of interest for North American mining hosts, especially in places with cheap, renewable energy like Texas. Other miners are eyeing locations closer to China like Kazakhstan or other destinations in Eastern Europe or Central Asia. 

Signs of the miner exodus became apparent on-chain May 18th when the amount of BTC sent by miners spiked to its highest level since March 2020. As news of impending crackdowns rolled in many miners likely transferred their BTC to either sell or self-custody, resulting in the spike.

Source: Coin Metrics Network Data Charts

This miner migration has caused Bitcoin’s hash rate to drop in the short-term. Hash rate dropped by 50% in Q2, dropping to its lowest level since late 2019. The following chart (created by @TakensTheorem) shows hash rate annotated with China regulatory events. Ethereum’s hash rate has also fallen as miners have left the country due to the uncertain regulatory climate.  

Source: TakensTheorem

Hash rate should eventually recover once miners start to power back up in their new locations. However it won’t happen overnight since it will take time to build and set up enough facilities to accommodate the sudden influx of new demand. 

Although hash rate is down the Bitcoin network was designed to be able to withstand sudden drops in hash rate. Bitcoin difficulty automatically adjusts every two weeks to keep producing Bitcoin blocks at a target of every 10 minutes. Bitcoin difficulty decreased by about 28% on July 3rd, which is the largest downward difficulty adjustment in Bitcoin’s history.

Source: Coin Metrics Network Data Charts

Over the long-term this mass migration should be largely beneficial as it will help Bitcoin hash rate get further distributed around the world, and remove the previous concentration in China. It could also help improve Bitcoin’s environmental impact since miners in some regions of China relied on coal.

Changing World

The migration out of China of both miners and investors will significantly reshape the global dynamics of bitcoin but it’s not the only place in the world where the relationship with BTC is changing. 

In June the El Salvador congress passed a bill officially making bitcoin legal tender. A pivotal moment for bitcoin and crypto at large, El Salvador is now the first country to recognize BTC as a satisfactory form of payment for any form of monetary debt. While it will take some time for El Salvador’s experiment to play out, bitcoin adoption around the world is undergoing one of its largest transformations to date.  There will undoubtedly be more pushback and challenges as El Salvador begins to adopt BTC on a larger scale. But if El Salvador’s experiment goes well, this could be a turning point for large-scale BTC adoption around the world. 

Coin Metrics Updates

Q2 was also a big quarter for Coin Metrics:

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 110

Wednesday, July 7th, 2021

Get the best data-driven crypto insights and analysis every week:

Ethereum Gas Update

By Nate Maddrey and the Coin Metrics Team

Ethereum gas price has dropped down to its lowest levels since March, 2020. While a few months ago average gas price regularly reached 150-200 GWEI, average gas price has decreased to the 15-30 GWEI range since the end of May.

Source: Coin Metrics Formula Builder

Although it may seem like the drop in gas price corresponded with ETH’s price crash, gas price actually started trending downward in late April, well before the crash. Average gas price reached as low as 40 GWEI on May 1st.

Gas then temporarily spiked back up to 250-300 GWEI on May 12th and 19th, two of the worst days of the crash. But upon closer examination the large price spikes were caused by extreme circumstances and were relatively short lived. 

Around 11:00 UTC on May 19th, average gas price suddenly spiked from under 100 GWEI to over 2,000 GWEI in less than two hours. It then dropped down to about 300 GWEI a few hours later. 

Source: Coin Metrics Network Data Pro

The extreme gas spike occurred during one of the worst flash crashes in ETH’s history, as price dropped from about $3,400 to under $1,900. The sudden crash led to a DeFi liquidation spiral, which caused escalating gas prices as investors desperately tried to avoid liquidation. According to a report from AAVE, May 19th was the largest single-day liquidation total in the history of AAVE and Compound.    

Source: Coin Metrics Network Data Pro

But outside of the relatively extreme circumstances of May 12th and May 19th, gas prices have mostly been trending downwards.

Several factors have combined to contribute to the drop in average gas prices. First, the Ethereum gas limit was raised to about 15M on April 22nd. The gas limit raise allows more operations to fit into each block, and has helped ease congestion. 

Source: Coin Metrics Formula Builder

Additionally, Ethereum scalability solutions have started to take off since late April. Polygon, a sidechain scalability protocol for Ethereum, has gained momentum over the last few months, including adoption by AAVE and other DeFi protocols. Arbitrum, which uses optimistic rollups for scalability, launched in late May. More and more transactions are moving over to these scalability solutions, which helps remove more congestion from Ethereum and further ease gas prices.

Lastly, flashbots is helping to move DeFi arbitrage bots off the Ethereum blockchain. With the rise of decentralized exchanges (DEXs) like Uniswap, arbitrage bots have been a major contributor to high gas prices. Since DEX trades are viewable in the mempool and on-chain, bots will monitor incoming trades and then try to front-run them for arbitrage or other profit making opportunities. The timing of these types of trades are critical so these bots are typically willing to pay extremely high gas prices to try to outbid each other and get their transaction confirmed first. But flashbots has started to move this bidding process to a parallel chain, off of the Ethereum main chain. This has helped to reduce gas wars on ETH and bring down the overall gas price.    

Lower gas prices have contributed to a drop in overall ETH fees over the last two months. But with EIP-1559 set to go live in early August, there’s another potential consequence of lower gas prices: less transaction fees burned after the onset of EIP-1559. 

The following chart shows what ETH’s estimated annual inflation rate (30-day average) would look like if EIP-1559 results in 75% of fees being burned. There’s no way to know exactly what percent of fees will be burned compared to what percent will go towards tips once EIP-1559 goes live - this is just a historical estimate calculated by subtracting 75% of daily fees from the amount of ETH issued daily.

While estimated annual inflation rate would have dropped to less than 2% while there were high fees in March and early April, it’s now increased to above 3.5% following the fee drop in May. This would still be significantly below ETH’s current annual inflation, but it would not be as low as previously estimated. However, fees will likely continue to fluctuate over time. If the amount of total fees eventually goes back up ETH’s annual inflation will decrease thanks to EIP-1559’s fee burn mechanism.

Source: Coin Metrics Formula Builder

For a detailed explanation of Ethereum’s gas mechanism and how it will change following EIP-1559 see the Ethereum Gas Report. And to follow the data used in this piece and explore our other on-chain metrics check out our free charting tool, formula builder, correlation tool, and mobile apps

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

BTC active addresses are up 11.6% week-over-week, while ETH active addresses dropped by 6.5%. BTC now once again has more daily active addresses than ETH after ETH briefly topped BTC last week. BTC usage was down last week as the average time between blocks spiked, which caused less blocks than normal to get added to the blockchain. The increase in block time was likely due to a drop in hash rate, which can lead to less blocks getting mined. BTC hash rate stabilized over the last week after dropping to its lowest level since 2019 on June 27th. 

Network Highlights

Bitcoin difficulty decreased by about 28% on July 3rd, which is the largest downward difficulty adjustment in Bitcoin’s history. Although it has undergone a relatively large drop in hash rate over the past month, Bitcoin was designed to be able to adjust to sudden changes and remain secure. 

Source: Coin Metrics Network Data Charts

The difficulty adjustment comes after the time between Bitcoin blocks briefly topped over 23 minutes last week, its highest level since 2010. Bitcoin’s average block time increased during June, likely because of the ongoing miner migration out of China due to new stricter regulations. 

Average block time has decreased to about 658 seconds since the difficulty adjustment, much closer to the 10 minute target. 

Source: Coin Metrics Network Data Charts

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • State of the Network was published on a Wednesday this week due to the observance of the July 4th holiday. We will move back to our regular publication time of Tuesday at 8AM EDT for next week’s issue.

  • We’re excited to announce the new Coin Metrics mobile app. View real-time cryptoasset pricing and relevant on-chain data in a single app!  Download for free here: https://coinmetrics.io/mobile-app/

As always, if you have any feedback or requests please let us know here.

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Loading more posts…