The Unpublished Ledger: Deconstructing Jiang Zhuor's Loss-Rate Thesis on Bitcoin

The Unpublished Ledger: Deconstructing Jiang Zhuor's Loss-Rate Thesis on Bitcoin

Jiang Zhuor did not publish his data. That is the most important fact in his latest public read on Bitcoin.

The founder of B.TOP, one of the oldest mining pools in the Asian ecosystem, recently issued a cyclical judgment on the market's next major move. He anchored that judgment on two measures: the sector's "loss rate" and the volatility regime. No definitions. No datasets. No time horizon. The brief, which circulated as an industry fast-news item with no attributed source, reads less like research and more like a boardroom summary that escaped the building.

In my line of work, an unreferenced number is an invitation to forensics.

I have spent a decade watching mining operators talk about price. They sit closer to the machinery than any sell-side analyst, and their cost models are the closest thing crypto has to a hard reference point. But proximity is not proof. When a pool founder cites a loss rate without telling you whose loss, over which window, and measured against which cost base, the claim is untestable as published.

This is a bear market. Survival matters more than gains, and readers need to know whether their assets are safe. They also need to know whether the man making the loud call has receipts. So let us audit this call the way I audited smart contracts in 2017: item by item, until the numbers either pass or break.

The Man and the Metric

Jiang Zhuor is not a random influencer. He built B.TOP into a significant pool during Bitcoin's Asian institutional phase, survived two full halving cycles, and his public commentary usually tracks the mining sector's pulse: power prices, machine shipments, difficulty momentum. When he says miners are bleeding, he has a private P&L table in front of him. That is real information. It is also his information.

The brief itself is information-poor. The only hard claims are that the loss rate is elevated, that realized volatility is low, and that these conditions historically resolve into a large move. The rest is framing. Chinese crypto media routinely compress founder commentary into news without requiring methodology, so the reconstruction work falls to the reader.

Consider the phrase "loss rate." On-chain analysis offers at least three candidate definitions.

First, miner operating loss: the share of active hash rate whose all-in cost exceeds the value of freshly mined coins. Calculated from hashprice — revenue per terahash per day — against hardware efficiency and power cost. This is the definition a pool operator uses naturally, because it is the one he feels first.

Second, unrealized supply loss: the percentage of total supply held in UTXOs whose acquisition price exceeds the current price. Derived from MVRV-style value-band analysis. This covers the whole market, not just miners.

Third, the spent-output profit ratio, or SOPR: the ratio at which coins actually moving on-chain are sold relative to their prior acquisition price. Below 1, realized losses exceed realized gains in that window.

These three senses are not exchangeable. They peak and trough at different moments in the cycle. When a high-profile caller says "loss rate" without a definition, he is either being intentionally vague or speaking in-house shorthand. Both are red flags.

The market context sharpens the stakes. The halving has cut block subsidies. Difficulty sits near all-time highs. Hashprice is depressed enough that marginal ASIC fleets — the older generation — are operating at or below break-even. Volatility has compressed to levels that scream coiled. In this environment, a man in Jiang's seat tends to call a bottom. It is the most bullish thing a miner can say while his machines are running red.

Structure dictates survival in the digital wild. If we want to know whether the structure is sound, we have to open the ledger. Ledger lines bleed, but the arithmetic never lies.

Opening the Ledger

Let me begin with method. My forensic workflow is the same one I built during my early years as a contract auditor: define the variable, measure it on-chain, backtest its historical behavior, then try to break the conclusion. I do not accept momentum narratives as data. Every transaction leaves a ghost in the hash; the analyst's job is to interrogate that ghost.

Loss rate, measured properly

Start with the miner's ledger. Hashprice — expected daily USD revenue per terahash — is the first line. Multiply by a machine's hash rate and you get gross revenue before power and amortization.

A legacy S19-class unit runs roughly 100 terahash and consumes about 3.1 kilowatts. At current bear-phase hashprice levels, that machine grosses somewhere in the single digits per day in USD. Subtract power at an industrial rate of $0.06–$0.08 per kilowatt-hour, and the machine lands at a thin margin or a small loss, depending on hosting, maintenance and debt service. Newer S21-generation machines run near 15 joules per terahash versus the S19's 30, so they stay profitable longer. The distribution of hardware efficiency is the key variable.

Hashprice itself has fallen by roughly 70% from the cycle peak, yet total network hashrate keeps climbing. That divergence — price down, computation up — tells you the marginal producer is becoming more efficient, not that losses are disappearing. The industry is replacing old units with new ones, bidding up difficulty even as revenue per terahash collapses. The loss rate is therefore concentrated in the tail of the hardware distribution: the S19-generation operators with expensive power contracts. That is a meaningful difference from earlier cycles, when the entire fleet bled together.

The analyst's move is to find the hashprice at which the marginal unit goes offline. That is the true "loss rate" line. My own model, built after the 2024 ETF data integration work in which I standardized Glassnode and CryptoQuant ingestion into a single pipeline, tracks hashprice against difficulty momentum. When hashprice stagnates near marginal cost for weeks, the network is testing its floor.

The on-chain signature of that stress is the Puell Multiple — daily miner revenue divided by its 365-day average. Historically, a print below 0.5 marks capitulation zones. But the level is not the edge; the duration is. In December 2018, the multiple reached roughly 0.3 and stayed depressed for weeks; the true bottom arrived later, after the aggregate loss had become unbearable. In 2022, Puell crossed below 0.5 three separate times, and each crossing was early. The final capitulation followed the FTX collapse, not the mining distress itself. I lived through that period with a different microscope. When Terra collapsed in May 2022, I was running emergency liquidity stress tests across 10 major DeFi protocols from a desk in Jakarta, watching correlated stablecoin exposure converge on a single risk factor. The lesson that stuck applies to mining as directly as to lending: a distress signal can stay red for months, and the market does not bottom when the signal starts — it bottoms when the signal ends.

Hash ribbons add a second layer. When the 30-day moving average of network hash rate crosses below the 60-day moving average, miners are switching machines off. In 2018 and again in 2022, prolonged ribbon compression accompanied the cycle lows. But a positive ribbon — hash rate accelerating — has historically been a better early-cycle signal than any loss-rate read. A rising ribbon after a negative stretch means the weakest producers have been replaced, and the remaining fleet is profitable at the margin. That inflection is visible only in hindsight, which is why I treat ribbons as confirmation, not prophecy.

So if Jiang is measuring miner operating loss, the metric is real but it is not a timing oracle. It is a condition, not a direction.

Loss rate, as the market feels it

The second definition, unrealized supply loss, classifies every coin by age band and realized price. In the current regime, the dominant losses sit with short-term holders: coins acquired in the last one to six months. Long-term holders, by construction, bought earlier; their cost basis is below price. That is actually a healthier bear-market structure than most people realize. The realized losses of long-term holder cohorts are historically far below the peaks reached in 2015 and 2018, which suggests that the cycle's forced selling pressure is coming from a narrower slice of the supply.

The third definition, SOPR, can be monitored at daily and weekly granularity. A sustained print below 1 tells us selling pressure is coming from loss-making hands. A spike in spent outputs aged three to six months, moved at a loss, is the classic "weak hand removal" event. If that is what Jiang is watching, the honest read is that we are inside a distribution phase where supply is transferring from weak to strong hands. That is bullish on a calendar-year basis. It provides no weekly instruction.

Which "loss rate" did he mean? The brief does not say. My best-confidence inference is that he means a blend: his pool's operating margin, which degrades first and recovers first, plus the market's realized-loss behavior around him. He does not need to define it for his audience, because his audience is his client base. They already know his cost structure.

Volatility compression, measured honestly

The second anchor, low volatility, is easier to verify. Thirty-day realized volatility has been declining for weeks, and the Bollinger Band width on the daily chart sits in its lower historical percentile. This is not a contested observation; it is mechanical fact. The argument that compression precedes expansion is also statistically sound, because volatility is mean-reverting. A rubber band that is squeezed inward has to snap outward; this is finite variance working as designed.

But here I must slow the tape. The chain tells us the spring is wound. The chain does not tell us which way the spring uncoils.

In the first quarter of 2020, volatility collapsed for exactly the same mechanical reasons. The pandemic shock resolved the compression downward, and Bitcoin lost roughly half its value in days. Then the same metric regime resolved upward in an equally violent rally. In late 2023, a prolonged low-vol grind broke to the upside and carried into 2024. The universal trait is expansion; the direction was a coin flip with a slight upward drift over longer horizons. If Jiang's thesis is only that a big move is coming, he is not wrong — but he is not useful. If he is adding a directional view, he is doing so with data he has not shared.

The derivatives market confirms the ambiguity. The 25-delta risk reversal in the options chain is not flashing the extreme put skew or call froth that marks a directional setup. Positioning is light, funding is flat, and implied volatility sits low relative to its own history. The futures basis is near zero, and the term structure is in a shallow contango that implies the market is paying almost nothing for convexity. Participants are not leaning; they are waiting. In the institutional framework I helped build — the one that cut our data latency from hours to seconds — waiting is not a position. It is an invitation to be wrong in the right direction.

The evidence chain, assembled

Let me compile what the chain actually confirms today.

First, miner reserves are near cycle lows. The aggregate miner wallet is lean, and the sell-side overhang from the 2021–2022 distribution has largely been worked off. There is no evidence of a miner dump of the kind that marked previous capitulations.

Second, exchange balances are low and flat. Bitcoin held on centralized venues sits at levels that historically precede structural scarcity, and netflows are not showing panic. Stablecoin reserves on exchanges, meanwhile, have been quietly accumulating. Without stablecoin inflow, no rally has legs; the current slow build is a dry-powder signal that institutions are positioning for entry, not exit.

Third, long-term holder realized price is far below market price. The largest holding cohort is not in distress. Losses are concentrated precisely where bear-market endings are usually concentrated: the late-arriving weak hands.

Fourth, leverage is restrained. Estimated leverage is nowhere near the extremes that preceded the 2021 top, so a move, when it comes, will carry less liquidation-driven acceleration.

Assemble the chain in sequence: declining miner reserves, low exchange supply, contained losses, flat volatility, neutral positioning. The structure is consistent with the late-accumulation phase of a bear cycle. If Jiang sees the same chain, he sees all of this. But "consistent with a bottom" and "the bottom is here" are different propositions, separated by weeks or months of possible discomfort.

The core insight is that the loss-rate signal is a confirmation tool, not a leading indicator. It tells you the cycle has matured; it does not tell you the final capitulation has printed. Volatility compression is a precondition for a large move, not proof of its direction. The combination of the two is a loud reason to prepare, but not a license to pre-position.

What he knows that I do not

There is one legitimate edge Jiang holds that no public analyst can fully replicate. He runs a mining pool. Inside that pool runs a stream of order flow the chain does not cleanly expose: clients' incoming ASIC deposits, power prepayments, and the choice between selling coins on OTC desks or moving them to exchanges. That telemetry is a genuine lead over public hash rate data. If he is reading his client book and seeing deferred sell orders, expanding power reserves, or new machine deployments, he has a real basis for a bullish call. He simply does not publish that basis.

As a hedge fund analyst, I have learned not to despise that asymmetry; I have learned to price it. A founder's private telemetry is worth more than any single dashboard I subscribe to. But the moment he converts private telemetry into a public price call, he has crossed from data into narrative. My job is to separate the two. His loss-rate line is narrative until he shows receipts.

This is the lesson I carried out of the 2017 ICO audit era: facts are not facts until they are anchored to a specific ledger line. I spent four months reviewing ERC-20 contracts, and the projects that leaked credibility were the ones that explained everything at the level of "trust us." The projects that held up provided a checklist: supply, vesting, tests, invariants. Jiang presented the equivalent of an unaudited clause. The arithmetic might be right. The disclosure is still a flaw.

There is also a structural shift that mining-based models have not fully absorbed. Since the ETF approvals in 2024, institutional order flow has added a layer that the old mining framework was never designed to capture: ETF issuance and redemption, custody flows, and basis trades by arbitrage desks. The mining loss rate is no longer the center of gravity it was in 2017 through 2022. Miners are still marginal sellers in stressed markets, but they are no longer the marginal buyer in recoveries. That role has passed to the ETF custody complex and the macro desks behind it. Jiang's framework is built for an older market; the chain has moved on.

The Counter-Thesis

Now the uncomfortable part: the counter-thesis.

Miners talk their books. When a pool operator's fleet is underwater, a public thesis that the bottom is near is not always an analysis of the market; it can be a margin call disguised as insight. The connection between private interest and public argument is not fraud — it is bias. And bias in a founder is precisely the variable that data must be able to override. The loss-rate metric he cites could be measuring his own operational reality: the average efficiency of his fleet, his power contracts, his financing costs. None of that randomness generalizes to the market.

The empirical record punishes early miners. In 2018, the mining sector was underwater for months. The Puell Multiple printed capitulation levels repeatedly, and each "this is the bottom" call was followed by another leg down. The December bottom arrived long after the loss rate was already grotesque. A miner who positioned on the loss rate alone in October 2018 was down another 40% by the turn of the year. The metric was true; the timing was tragic. In 2022, the picture differed: the final capitulation came fast, inside a few weeks after FTX, rather than as a long decay. The lesson is that the loss-rate time series is path-dependent. The level matters less than the transition, and the transition is visible only in hindsight.

Then there is the tautology problem. Volatility compression preceding expansion is a statistical identity for mean-reverting processes. It carries no directional content and no reliable timing. In 2019, the squeeze resolved south by about 30% before resolving north by 100%. Anyone who used "low vol means a breakout is coming" as a buy signal was caught in the first leg. "The move will be large" is different from "the move will be up" and very different from "the move starts next week."

The correlation-versus-causation trap is the through-line. Miner losses, exchange supply and realized volatility are all co-moving states of the same system. They are outputs of the same economic pressure, not independent causes of price. If BTC falls, miner revenue falls, the loss rate rises, and miners sell more to cover power bills — which pushes BTC lower. This reflexivity means that citing the loss rate as a predictor resembles a doctor citing a patient's fever as a predictor of recovery. The fever is a symptom. The loss rate is a symptom. The infection is the cycle, and the cycle's duration is the one variable the metric cannot forecast.

That is the portion of Jiang's call that deserves a cold stare. Not the arithmetic. The narrative architecture around it.

The Week Ahead

Over the next week, I will be watching three signals that turn the thesis into a testable claim.

First, the realized-loss channel. I am tracking the three-to-six-month spent-output cohort. If SOPR in that band spikes below 0.98 on rising volume, a fresh capitulation leg is underway and the loss-rate thesis has not finished printing. If it holds above that line, the bleed is contained.

Second, the difficulty directory. The next adjustment will tell us whether hash rate is contracting. A negative adjustment beyond 3% confirms marginal miners are leaving; a positive adjustment means the strongest hands are running through the pain. That distinction is the closest thing we have to a leading on-chain vote.

Third, miner-to-exchange flows. The miner-labeled wallets on our internal feed — the same feed that cut our latency from hours to seconds in 2024 — will show intent. Seven consecutive days of miner-to-exchange inflows would violate Jiang's thesis. A sustained outflow, or a steady trickle toward OTC desks, would echo his call.

The compression will resolve. The question is whether the resolution rewards the patient or the premature. Do not pre-position on low volatility. Position when the expansion shows volume on the side it chooses.

The chain remembers what the founders forget. Jiang forgot to show his work; the ledger will not. Watch the hash, watch the exits, and let the arithmetic hand down the verdict.