Good Questions

Source

< On November 21^st Bryce Elder posed Five questions from an ignorant no-coiner about the crypto crash. Each of his five questions identified some interesting apparent anomalies.

Below the fold I look into each of his questions, asking how anomalous its anomalies really were and whether they have persisted into the New Year.

TL;DR none of them are really surprising but reaching that conclusion took a good deal of research.

Why doesn’t network difficulty fall?

Elder explains Bitcoin's difficulty mechanism:

Bitcoin’s proof-of-work algorithm has a difficulty ratchet to keep production steady. Network difficulty adjusts every two weeks, approximately, based on whether miners have been finding new blocks for the blockchain too quickly or too slowly.

And then identifies the first anomaly:

Difficulty tightens when a higher bitcoin price is encouraging miners to join the market, that much makes sense. But when a lower bitcoin price is squeezing margins and pushing the least efficient miners out, shouldn’t difficulty fall?

Here is the chart Elder is looking at.

Source

Then Elder asks:

Maybe difficulty lags behind the price? As said above, the ratchet only adjusts approximately every two weeks. Total hashrate offers a more dynamic measure of power deployed to the network, albeit an estimated one based on how long it’s taking to mine a block. Any change in market structure might show up there first.

And has it? Not really:

The first thing to note is that, depending upon power costs and where they are in the queue for the latest rigs from Bitmain, miners margins vary greatly. So Elder is right that, if one of the two-week adjustemnts increases the difficulty, some of the least economic rigs should be switched off. In theory, if the adjustment decreases the difficulty, some of these idled rigs should be switched back on.

Eyeballing the graph, between early July and late October the price was between $110K and $120K, a 9% range. During that period the difficulty increased from around 140trn to around 155trn, about an 11% range. These aren't big changes, but it appears a bit strange that roughly flat price coincides with a steady-ish increase in difficulty.

Elder has some possible explanations:

Do miners with sunk costs keep running on negative margins in the hope of getting lucky? Are a handful of big miners, maybe advantaged by free power or whatever, keeping difficulty high to drive out competitors, either inadvertently or as part of some devious plan to centralise production and control the network? Or has mining become 55 per cent more efficient since last November?

I estimate that during that time Bitmain and its competitors shipped an additional 1000EH/s to the miners, over and above the new rigs replacing obsolete ones. These leading-edge rigs would have been turned on immediately. During that time the hash rate rose from about 870EH/s to around 1120EH/s, or about 250EH/s. So around 150EH/s must have represented idled, less-efficient rigs being turned on. At the start of the period at most 85% of the rig fleet was working. At the end of the period, as the price started to fall, the most efficient miners had about 100EH/s more than they did at the beginning.

In the subsequent 2 months the price dropped from about $120K to around $85K and the hash rate dropped from around 1120EH/s to around 1060EH/s, or by 60EH/s. But in that time there ws an incremental 50EH/s of new rigs. So 110EH/s of the 150EH/s from the originally idle fleet of uneconomic rigs were turned off. This within the margin or error of my back-of-the-envelope estimates.

My take is that what happened was what should have been expected. Elder's view was too simplistic, and he was misled by (a) the noise in the graph, and ((b) the fact that the Y axis doesn't start at zero, making the noise much more evident.

Source

Elder goes on to note an issue I've been writing about for really a long time:

Related, is it a worry that just three mining pools accounted for more than 45 per cent of this week’s block production? Given nearly all the hardware used across the network is Chinese-made, with Beijing-based Bitmain Technologies alone having an estimated market share of 82 per cent, when do concentration levels among a few organisations become a security concern?

Yes, this concentration has been a problem for Bitcoin's entire history, but everyone has decided to ignore it. As I write, on 6^th January 2026, over the last three days Foundry USA and AntPool have controlled 51.2% of the hash power. As I explained at length in Sabotaging Bitcoin, there are practical difficulties facing an insider attack with 30% of the hash power, but with 51.2% things are much easier.

And, in November, Reuters reports that Bitcoin mining in China rebounds, defying 2021 ban:

Bitcoin mining is quietly staging a comeback in China despite being banned four years ago, as individual and corporate miners exploit cheap electricity and a data center boom in some energy-rich provinces, according to miners and industry data.

China had been the world's biggest crypto mining country until Beijing banned all cryptocurrency trading and mining in 2021, citing threats to the country's financial stability and energy conservation.

After having seen its global bitcoin mining market share slump to zero as a result of the ban, China crept back to third place with a 14% share at the end of October, according to Hashrate Index, which tracks bitcoin mining activities.

This may be what the 80% of new data centers in China that were empty because unsuitable for AI are being used for.

What’s the deal with Tether?

Zeke Faux described Tether as "practically quilted out of red flags" and they are still flapping. What attracted Elder's attention was this:

Stablecoins have long been pitched as crypto’s on-ramp. Swapping fiat money for a fiat-pegged stablecoin like Tether’s USDT or Circle’s USDC allows a trader to switch in and out of positions without having to touch tradfi.

Shouldn’t an on-ramp also work as an off-ramp? There’s not much evidence it does. In the six weeks or so when $1.2tn in value was drawn down from the cryptoverse, the market cap of USDT has increased by approximately $20bn:

Source

This USDT graph is updated from the one Elder used. It shows a clear break in the upward trend on 24^th October. In the 10 preceding weeks it increased by $17B (~10%) as Bitcoin traded beteween a low of $108K and a high on 6^th October of almost $125K.

In the 10 weeks since it has increased by only $3B as Bitcoin dropped from $111K to a low of $84K on 22^nd November, subsequently trading in a range from there to $94K.

The only other stablecoin that matters is USDC:

USDC hasn’t been quite as resilient but over the period is still basically flat:

Source

Here is its updated graph. On 24^th October its market cap was around $76B, it is now around $75B. So, yes, Elder was right, it has been basically flat for 10 weeks where the preceding 10 weeks it had increased by $8B (~12%).

Had the trend of the previous 10 weeks continued, the market cap of (USDT+USDC) would have been $25B higher than it is. One way of looking at this would be that traders "withdrew" $25B or around 10% of the (USDT+USDC) market cap on 24^th October. On this basis Elder is just wrong, the off-ramp has been quite busy.

Elder notes that:

Even if we assume a large percentage of stablecoins are used for non-crypto things (sports betting, remittances, crimes), the recent issuance still looks at odds with the trend.

First, this contradicts the premise of Elder's question as I understand it. Second, most of the uses Elder cites involve a chain of transactions from fiat to stablecoin to fiat on one or more exchanges. These would not cause a net increase in demand for the stablecoin, which would come from and go back to the exchange's reserves.

Then Elder suggest that:

Maybe demand is high because crypto traders have been parking money rather than seeking to withdraw it?

First, demand isn't high relative to the historic trend, it is now $25B lower. Second, even if we accept Elder's $20B number, this is peanuts relative to the $1.2T drop in aggregate cryptocurrency market cap.

Source

Finally in this section Elder asks:

More idle money in the system ought to be good news for the likes of Coinbase, which uses the promise of higher yields on USDC deposits to sell monthly subscription schemes. And how have Coinbase shares been doing?

The day Bitcoin hit the skids on 8^th, COIN closed at $387.27. By 11/20 it was down 38% while Bitcoin was down 47%. It can hardly be a surprise that COIN is highly correlated with Bitcoin.

What explains trash crash PTSD?

Next, Elder asks why cryptocurrencies haven't resumed their progress moonwards:

A popular argument among crypto commentators is that token prices are down because traders are still digesting one bad day in early October.

He suggests some reasons:

Reasons for the October drawdown go from banal (maybe the high-beta cryptosphere just amplified an equities pullback on US-China trade tensions?) to wonky (maybe it all cascaded from a weird synthetic stablecoin depegging on one marketplace?) to the darkly cynical (maybe the big sharp drop was to let bucket-shop crypto brokers close out customer positions they’d never actually bought?).

Source

In anything to do with cryptocurrency, and definitely in this case, I'd go with "darkly cynical". He then asks:

If a market’s not deep, efficient or clean enough to digest a bit of one-day volatility, why get involved?

I'd agree with Elder on this. If we look at the log plot of Bitcoin's history, we can barely see the 47% drop in 6 weeks last October. We can just see, to take some recent examples, the 48% drop in 5 weeks in early 2020, and the 42% drop in 2 weeks in mid-2021. Traders who don't learn from history are doomed to repeat it. Of course, the volatility is precisely the thing that brings the traders to Bitcoin.

Where’s the volume?

Elder notes a wave of selling:

Crypto exchange-traded products have been haemorrhaging money all week. Spot ETP net redemptions yesterday were $1.14bn, including $901mn just from bitcoin ETPs, according to JPMorgan estimates. That’s the worst single day for net outflows since February.

With so much selling, you might expect to see an increase in bitcoin velocity, which measures the rate at which tokens move on the chain.

Source

This graph covers a longer history than Elder's. Already back in 2021 Igor Makarov & Antoinette Schoar found that the "rate at which tokens move on the chain" was irrelevant:

90% of transaction volume on the Bitcoin blockchain is not tied to economically meaningful activities but is the byproduct of the Bitcoin protocol design as well as the preference of many participants for anonymity ... exchanges play a central role in the Bitcoin system. They explain 75% of real Bitcoin volume.

Velocity dropped sharply until November 2023, and then continued to drop gradually. November 2023 was the start of a major increase in Bitcoin's price, which coincided with a sustained increase in trading volume on exchanges.

Source

It thus seems likely that the trend identified by Makarov & Schoar that on-chain activity was largely confided to exchanges increased between 2021 and 2023, and was then saturated. Elder's explanation is only part of the story:

Bitcoin velocity has been plummeting for years, for reasonable reasons. “Digital gold” overtook “internet money” as the preferred reason to hold, while derivatives like perpetual futures removed any need to faff around with the underlying asset.

It isn't just derivatives. Spot trading happens on exchanges. The blockchain is pretty much only used for inter-exchange netting transactions. So Elder's question misses the point:

Nevertheless, is it odd to have a sudden wave of selling that’s almost invisible in the underlying asset? Bitcoin velocity has barely changed over the past month, having bounced meekly off a record low in early October. Why?

Having a vast derivative market based off a much smaller spot market on exchanges based on a tiny set of transactions on the blockchain is a wonderful playground for traders, because it is easy to manipulate.

Is past performance indicative of future results?

Source

Spoiler: No. In this section Elder expresses skepticism about a piece by Alex “Crypto Alex” Saunders, of Citigroup suggesting that:

The halving cycle is a reason that long-time Bitcoin holders are nervous. We show the price performance in the years after halvings in Figure 3 with the second year showing weakness. These crypto winters have been associated with 80%+drawdowns in the past as shown in Figure 4.

Elder doesn't need any help from me on this "chart necromancy". Halvenings primarily affect miners by roughly halving their income without a corresponding decrease in costs. This may force them to sell coins they have stashed, which at the margin may drive the price down. But with miners' income currently around $40M/day and recent volume peaks on major exchanges around $1.4B/day this is likely to have marginal impact.