Tutorial 3 - Fundamentals of Proof-of-Work Mining
Bitcoin mining is one of the most interesting, but least understood parts of the network. This session will explore the fundamentals of mining dynamics, and how we track them with on-chain metrics.
Bitcoin’s security is assured by miners, whom are paid in BTC to provide computational power, compile transactions into blocks, and build the blockchain. The total computational power provided by all miners on the network is estimated by a metric called Hashrate, and the complexity of the puzzle they must solve is called the Mining Difficulty. The amount of hashrate on the network is highly dependent on the market price of BTC (Miner Revenue). As a result, the mining industry also experiences market bull and bear cycles. Mining hardware also becomes increasingly efficient over time, transitioning from CPUs --> GPUs --> FPGAs --> ASICs. We must therefore observe many mining metrics in both linear (left, blue) and logarithmic (right, orange) scales.
In this section, we will explore the fundamentals of Proof-of-work mining including:
- The relation between Block-time, Difficulty, and Hash-rate
- Assessing mining rewards and revenue
- Models mapping out Miner market cycles
The Mining Difficulty can be though of as the complexity of the puzzle miners must solve to find the next valid block. It is a self-regulating algorithm that adjusts the puzzle complexity (called the Difficulty Adjustment) depending on how fast or slow blocks are being mined. The Mining Difficulty will change once every 2016 blocks (approx. 14-days), to achievce the following goals and objectives:
- When more miners are coming online blocks are found faster than 10mins, and the Difficulty will increase (harder puzzle).
- When miners are leaving the network blocks are found slower than 10mins, and the Difficulty will decrease (easier puzzle).
- This constant re-targeting of a 10min Block Interval makes sure that no matter how much mining power is online, the Block Interval will be consistent, and the pre-defined coin supply issuance will remain the same.
Every time a miner finds a new block, they encode the Timestamp into the block header. The Bitcoin protocol measures the average time between blocks in a 2016 block difficulty window, and then performs the Difficulty Adjustment:
- Difficulty needs to increase if the average block interval is faster than 10mins.
- Difficulty needs to decrease if the average block interval is slower than 10mins.
The chart below shows the Mean Block Interval (orange), its 14-day Average (black), and the 10min target (600 second, dark blue). We can see that Mean Block Interval is lower than the target for most of Bitcoins history. This is a result of the expansion, and efficiency growth of the mining industry leading to faster blocks, and thus increasing Difficulty. However, There are some instances where the Mean Block Interval is higher than the target for an extended period, usually in bearish markets or during shock events (e.g. May 2021). This signals that miners are leaving the network, and blocks are slowing down.
Hashrate is the most widely known mining metric. It measures the estimated number of SHA256 computations performed by all miners every second. Hashrate has units of hashes per second (H/s), however it is often quoted in Terrahash/s (TH/s, one thousand-billion hashes), or more recently in Exahashes/s (EH/s, one billion-billion hashes).
For a sense of scale, the hashrate of 220 EH/s (as of May 2022) is equivalent to all 7.9 Billion people on earth, guessing a SHA256 hash 27.85 Billion times every second!
Note: Hashrate is actually an estimated metric. We know the puzzle Difficulty, and we can see the actual block timestamps, from which we compute the Block Interval. We then use these two metrics to estimate an implied hashrate to produce this result.
💡 Hint: Hashrate has quite a lot of natural variability as a result of the probabilistic nature of block solving, difficulty adjustments, and changes in actual mining hardware. It is recommended to apply a 14-day moving average, or median to smooth out this variability to observe the underlying trends, and quote more accurate hashrate values.
Miners are paid the Total Block Reward by the protocol, with revenue coming from two sources: the Block Subsidy, and the Transaction Fees. Both components are denominated in BTC, however we can easily calculate their USD value by multiplying the BTC block reward by the price when they are received.
The total USD denominated transaction fees paid to miners by Bitcoin network users every day. Note how historically fees tend to spike with increasing market attention (bull markets) as a result of more users, more network congestion, and greater urgency in getting transactions confirmed. Conversely, transaction fees have historically declined during quieter periods when there is less network congestion, typically seen in more bearish markets.
The total amount of BTC paid in aggregate to miners each day, calculated as sum of the Block Subsidy (issuance, newly minted coins) and Transaction fees.
We can see that this metric declines over time as a result of the Halving Events which reduces the Block Subsidy issuance by 50% every 210,000 blocks (approx. 4yrs)
If we multiply Total Block Reward [BTC] in the chart above by the BTC/USD price on the day the reward is paid, we can calculate the USD denominated revenue paid to miners each day. We can use this metric to assess the Security Budget paid by the protocol to miners for network security. We can also see that over time, whilst the BTC Block Reward is falling (due to halving events), the USD denominated revenue has increased over time as a result of higher BTC/USD exchange rate. This dynamic also includes Transaction Fees paid, which tends to change as market cycle conditions change.
As noted above, miner revenue, and thus profitability, is directly linked to the BTC price. Mining is also an extremely competitive industry, requiring careful management of costs (power, hardware, labour etc), financing (debt, equity), and the timing of mining rig expansion/upgrades in line with favourable market cycles.
- During bullish markets miners have more fiat denominated income, and thus greater opportunity to invest in more or newer generation mining rigs. This leads to an increase in the applied hashrate, and thus an increasing Difficulty.
- During bearish markets miners have less fiat denominated income, but still have fixed costs to pay. During extended bears, this can lead to some miners shutting off some, or all of their rigs, either temporarily, or permanently. This leads to a decrease in the applied hashrate, and thus a decreasing Difficulty.
The Difficulty Ribbon is a powerful tool for monitoring these mining market cycles. The Difficulty Ribbon compares Mining Difficulty across multiple moving averages: 200d, 128d, 90d, 60d, 40d, 25d, 14d.
- Healthy Hashrate Expansion is signalled by the faster moving averages (e.g. 14d, 25d, 40d) rising faster than slower averages (90d, 129, 200d). This creates a Difficulty Ribbon expansion and signals a healthy and growing mining industry.
- Mining Market Contractions are signalled by the faster moving averages falling faster than the slower averages, and even dropping below them during periods of extreme financial stress. This creates a ribbon contraction and signals a weaker and stressed mining industry.
The charts below are shown in Linear (top) and logarithmic (bottom) scale to observe recent performance, and historical performance, respectively.
Bitcoin mining is one of the most interesting, but least understood parts of the network. This session will explore the fundamentals of mining dynamics, and how we track them with on-chain metrics.
Topics for Discussion:
- Understanding difficulty, block intervals and hashrate
- Assessing miner revenue and profitability
- Tracking miner capitulation cycles and behaviours
- Confluence between miner cyclical indicators.
These mining metrics are laying the foundation for understanding Proof-of-Work mining. With a Glassnode Advanced plan, we can observe the dynamics of the mining market to assess its health, and whether miners are in profit, or under income stress. Some example Advanced metrics include:
- Difficulty Ribbon Compression creates a cyclical oscillator out of the distance between the fast moving and slow moving Difficulty moving averages. This helps to spot where a difficulty ribbon inversion may be in effect.
- Miner Revenue per Exahash models the approximate USD income for each Exahash applied to the network. This can be a useful tool for evaluating a miners expected revenue in comparison to the market competition.
- Puell Multiple models the current daily aggregate miner income against their yearly average. During mining industry booms and busts, the Puell multiple will reach extreme highs, and lows, respectively.
- Mining Pulse is an oscillator tracking where miners are producing blocks faster (-ve) or slower (+ve) than the 600-second target block time. This can identify when stress or operation expansion is occurring within the mining industry.
- Hash Ribbons track phases of mining market boom and bust. During expansion, the 60DMA of hash-rate will rise faster than the 90D, whilst during capitulation it will fall below, signifying a loss of hash-rate online.
