A remarkable trend has been happening in the cryptocurrency mining scene over 2018. With the prices diving the whole year, the hashrate has finally started dropping as well.
Testing the axiom
Let us try to interpret the available data and test the axiom “hashrate follows price”. Could the opposite be true instead? Is there even any correlation?
Analyses we’ve seen so far focus either on creating derivatives (the best in my humble opinion being NVT and the smoothed NVT Signal) or misinterpret the raw data.
The reason for this is fairly straightforward when you think about it. Mining is most likely the most opaque part of the cryptocurrency scene. It is difficult to get proper data, financials, statements, electricity prices, etc. Without experience running your own machines, you are left wondering about the intricacies. Even then you have to extrapolate based on your very limited experience. Oftentimes, publishing your views even indirectly damages your operation as you divulge data your competition would otherwise be unable to exploit, leaving you in a worse position than before. For example when you share your mining picks, you dilute your own share in the network and lower your rewards if they decide to join the network. When you share your know-how on of to run an operation more efficiently, the same thing happens.
This is one reason why most analysts ignore mining and instead focus on “tokenomics”, “generalized mining” (read = VC bought into a proof of stake coin and want to stake it). This creates false narratives like “mining death spiral” and other absurd buzzwords.
Let’s analyze the predicament of Proof-of-Work (POW) coins using the only reliable data point we have so far — hasrate. Hashrate (and difficulty) is the best raw miner sentiment data we have and it’s been criminally underutilized. It is the sole data point we can draw directly from the network as it is directly tied to the blockchain itself with no guesswork involved.
Litecoin
It might be surprising that I’m not starting with Bitcoin, but I’d like to cover Litecoin (LTC) instead. It’s the archetype of the axiom we’re discussing. This might be thanks to it being a fairly self-contained system.
First the facts: in the bear market of 2014/2015, LTC retraced 98% from $50 to $1.20 over about 18 months. This was accompanied by a lagging drop in hashrate, with the lowest hashrate low occurring almost three months after the price drop to $1.20.
An interesting observation: The hashrate kept going up during price decline until it topped out a year later in December 2014. Let’s not forget this means new machines, more electricity, larger or new facilities, networking — the whole shebang. The selling pressure and evaporating margins kept creating pressure on the market, which finally had a savage breakdown after the slow bleed. The newest and weakest miners were leaving in droves, while older miners were realizing losses on their electricity prices. The hashrate first disregarded and then followed price in the bear market.
Throughout what was considered the LTC accumulation, the hashrate ranged between 1 and 1.7 terrahash per second (TH/s). A nice 50% range. The range coincided with the 3 to 5 dollar range in LTC prices for about two years. At this point, the industry standard in Scrypt mining was the KnC Titan, which suffered from constrained supply. In fact, that might be why the hashrate ranged — new miners did not want to join the network unless on the most efficient machines, which were almost impossible to purchase.
Next, observe the gentle hashrate breakout in November 2016, breaching 2 terrahash for the first time in history. Another data point relevant here is the halving, which, thus far, has occurred only once.
This was not visible in price at first. The price first disregarded and then followed hashrate in the bull market. The reason for the hash increase became apparent later when the Bitmain Antminer L3+ officially came out. It remains the scrypt industry standard two years later, for better or for worse. I‘m pouring gas into a fire here, however, I have only one plausible explanation for this, which is consistent with the behavior of this ASIC manufacturer.
Let’s assume the ASIC producers “test” their hardware before official release and that they test them at scale. This would imply a significant growth in hashrate since the new miners are more efficient than their older counterparts. Unlike the old miners, the new ones initially aren’t mining at breakevens nor at a loss. The producers are mining with the intent to sell these machines a couple of months later, which effectively means they can achieve a ROI on the machine even before it’s sold to the end customer.
The narrative is that the manufacturer was mining on these machines c. 6 months prior to the official release, which caused a hashrate breakout.The price followed.
Now let’s turn to the current situation and the ensuing speculation — history does not repeat itself, but it rhymes.
Litecoin has retraced 93% from $350 to $30 since December 2017. If LTC were to retrace 98% from the top, this will put it in single digit $ territory. Percentage retraces become very hazy with these high numbers and cannot be relied upon.
The LTC hashrate topped out at 340 TH/s in May 2018, c. half a year after the top in price. In November 2018, six months later we’ve had a savage drop in price, evaporating the already cutthroat miner margins. Most miners are now in the negatives. The ones that are not, have such low profits that they are considering whether to turn their machines off to wait out the bad times.
The hashrate now is 175 TH/s, which is a 50% drop compared to that in May 2018. This is very very similar to the 2015 drop, from 1.77 TH/s to 0.85 TH/s. Miner capitulation should follow the price capitulation as miners react to the price and turn off their machines over the next weeks.
Another point to consider is the cost of the miners themselves. In January 2018, an Antminer L3+ from the official Bitmain shop cost around $2000 (the price varied based on the batch), with a delivery period in April. At the time of writing, two fresh L3+’s go for the price of one PSU that supports it. That’s right, an L3+ costs fifty bucks. What a bargain. It would be interesting to compare the price drop of miners compared to the last bear market, however, it’s impossible. Historic miner prices are impossible to find online. The irony of blockchain miners relying on oral history, for lack of written history.
The miner scene is much more mature now than it was during last market cycle. However, the narrative remains the same. Retail miners turn off their machines as the price dives as they cannot afford to pay the bills. Eventually, only a handful of strongest hands remain. They either have ridiculously low operational costs (OPEX) or have saved enough money to support their operations. The hashrate then ranges at a fairly constant rate as the market is recovering for a year or two. There will be more people capitalizing on this situation this time around. They will consolidate and buy the miner blood over the next year as the price stabilizes and halving occurs. This is of course a speculation, however, the nature of the market is such that it learns and adapts over each cycle.
Bitcoin
In many regards, the industry standard, Bitcoin (BTC) retraced c.82% after the second 2013 bullrun, a drop from $1100 to $200 over about a year and a half. Hashrate, on the other hand, never properly dropped. There never happened a proper drop in hashrate, however, it did stop climbing at a massive rate during the whole of 2015 and the acceleration picked up again in November 2015.
Again let’s compare hashrate to price. After the price topped out, hashrate kept going up at a steady pace increasing by a factor of 2 in about two to three months) after the price topped out. This was followed by a lull where the difficulty of mining increased only by 30%. The hashrate first disregarded and then followed price in the bear market.
We can observe the acceleration in hashrate growth while price stayed steady. I would love to be able to point to more data to support the slowly developing hypothesis that the price first disregarded and then followed hashrate in the bull market. However, in this case, we can clearly see they both broke out at the same time — hashrate simply acted as confirmation on top of the price action itself.
The only data point we can debate here is the halving, which has so far acted as a catalyst for the bull markets. Unfortunately, we will see diminishing returns on this effect as more and more of the miner value will come from the transaction fees themselves rather than mining rewards. This has already happened during the top of the 2017 bubble.
You have probably noticed that there is an implied potential bottom near 3500$. The origin for this speculation is mining economics once again. With 80% miners already tapping renewable resources as their source of energy, the actual electricity prices will be below 0.05 $/kWh. The reason is that a Bitcoin miner can skip the distribution costs by tapping the renewable source directly. These sources are often far away from the general population, creating an abundance of electricity where it’s not needed. This is especially the case with what Christopher Bendiksen calls stranded/wasted hydro.
This suggests cash cost per 1 BTC of around 3400$ (what I call the breakeven in the article). If we include the CAPEX (all the hardware, building the actual infrastructure around the mine, networking, etc.) into the calculation, CoinShares came up with the figures below.
The prices we’re interested in hover below 6000$ in this case. This is fascinating: the miners are running their machines just to get a partial return on investment in their investment horizons. So is the 3000–3500$ mark where the situation grows critical? Not really. All fingers point towards this price being the equivalent of a Goldilocks zone, where retail miners are shaken out and machines are relocated to cheaper locations. The net effect is a stabilization in hashrate until there is a catalyst for the next price/hashrate increase.
The latest article by Bitmex Research gives a preliminary confirmation. The article provides an analysis of the daily revenue compared to electricity costs at 5 c/kWh as the benchmarks. We can see the price and hashrate stabilizing into a new equilibrium, with profit margins of c. 30%.
Market-wise, we’re finally shaking out miners both by forcing them to sell (the market has become saturated with ASIC’s) and to turn off the machines and hold on to them instead of dumping them on the market. The price of a regular ASIC has gone down by about 80% since the start of the year — one can buy used miners as low as 150$ with PSU included nowadays.
The hashrate is down 40% from 60 exahash/s (EH/s) to c. 35 EH/s at the time of writing. This is very similar to the LTC situation in 2014 than compared to BTC itself in 2015.
To put this into the hypothesis speak — hashrate is now finally following price, which would imply peak bear market, with stabilization imminent. Unfortunately we cannot draw any conclusion until we see a price stabilization, but the Litecoin analysis does give us some parallels to consider.
At a glance, the profitability of an average ASIC may seem fine, but one has to consider that most of the miner chips listed in the lower part of the table (industry standard) have a higher weight of the market than the rest of the market as a whole. The newer miners are mostly from the second half of 2018 with shipping dates of 3+ months away. Guess what happens with the miner in the meantime?!
There was also talk of Bitcoin cash hash wars stealing Bitcoin’s hashrate as the narrative for the hashrate drop. Consider the current hashrate for BCH, which is down to 1 EH/s, from a stable range of 4–5 EH/s. This is only a single digit percentage of the Bitcoin hashrate. Therefore we can conclude that the BCH split into BCH SV and BCH ABC did not have any significant impact and didn’t steal hashrate from the Bitcoin network.
Let’s reiterate . There is a larger group of very smart people in the market who survived 2014/2015. Now, in the next bear market, there will be an even larger group of people who will have survived 2018(2019?) bear market. Market participants learn. Mining is one of the freest markets out there. This means it is antifragile — when it gets wounded, it heals back stronger.
The conclusion from part 1
We have a simple problem that cannot be fixed — we don’t have enough data on the cryptocurrency market cycles. Now we are going through only the second proper Bitcoin bear market.
From what we have seen so far, hashrate follows price at the end of a bear market. Price follows hashrate at the beginning of a bull market. We will be testing this further in the next part of the article, where we will be looking at DASH and Ethereum.
I’d like to thank Brandon, CJuggz, and Notsofast for proofreading the article and throwing ideas at me.
