The number that landed in my inbox was 102. Not a price, not a yield — a ratio. Huobi HTX's October Merkle Tree Proof of Reserves placed the exchange's ETH backing at 102% of user liabilities. BTC sat at 104%. DOGE, 102%. A two-point cushion between solvency and the abyss, restated for the forty-eighth consecutive month.
I have spent nineteen years reading documents like this one, and I have learned to distrust numbers that arrive on schedule. A reserve ratio is not a promise. It is a photograph — and photographs lie by omission. What the snapshot does not show is the thing every auditor learns to hunt for first: the liabilities that were never placed on the tree.
This is where I want to begin, because the HTX report is being read by most of the market as a green light. I think it is better understood as a mirror.
Context
For anyone arriving late: Proof of Reserves became an industry ritual after November 2022, when FTX's collapse revealed that customer assets had never been segregated at all. In the panic that followed, exchanges rushed to publish Merkle Tree attestations — cryptographic structures that hash user balances into a tree and let each user verify their own leaf. HTX, formerly Huobi and founded in 2013, was among the earliest movers, and it now advertises forty-eight months of uninterrupted disclosure.
That history matters, but not for the reason HTX wants. Longevity of disclosure is not evidence of quality of disclosure. A protocol can publish the same weak instrument every month for four years and call the repetition integrity. The code remembers what the market forgets — and what the market has forgotten is that Merkle Tree PoR was always a floor, never a ceiling.
Core
Here is the methodological heart of it, and it is worth slowing down.
A Merkle Tree proves that the balances inside the tree exist and are backed. It cannot prove that liabilities outside the tree do not exist. Off-book borrowing, hidden debt, contingent obligations to lenders or market makers — none of this appears in a hash tree. The instrument is structurally blind to the exact failure mode that destroyed FTX. It guards one door while leaving the other unlocked.
Then there is the question of what the reserves actually are. Ownership is not custody, and custody is not freedom from encumbrance. Reserves can be temporarily borrowed, pledged elsewhere, or double-counted across counterparties. The industry has a name for this: window dressing. You rent the assets for the snapshot, return them the next morning, and the ratio never blinks.
Based on my own audit work — I spent six months inside Uniswap's early V1 contracts in 2017, tracing how the constant product formula quietly shifted trust from counterparty to code — I can tell you that the most dangerous vulnerabilities are never in the formula. They are in the assumptions the formula leaves unstated. HTX's report leaves several unstated at once.
The first is the auditor. The disclosure says the reserve information is "accurate, clear, and traceable." That is a self-assessment. No independent firm is named. When Binance publishes its reserves, it uses zk-SNARK proofs; when OKX does, it ships independent verification tooling. HTX ships a claim.
The second is granularity. Since April, HTX has folded USDT and USDC into a single category called "USDs." The report claims each asset is "individually recorded and verified" — which cannot be true if two assets are merged into one line. Aggregation is where transparency goes to die. If USDC reserves are short, a surplus of USDT can quietly cover the gap, and no reader would ever know. The two most important stablecoins in the market have been compressed into a single, un-splittable number.
The third is the cushion itself. BTC at 104%, ETH at 102%, DOGE at 102%. In a market where a single volatile week can move an asset ten percent, a two-to-four point buffer is not a buffer. It is a rounding error standing between an exchange and a bank run. And the buffer includes HTX's own platform token at 103% — an asset whose value is itself a function of the exchange's survival. Reading the silence between the blocks, you notice the circularity: the exchange is using its own credit to back its own promises.
Contrarian
The consensus reading is that HTX is doing the responsible thing — publishing monthly, staying above 100%, committing to instant withdrawals. I want to push against that.
The real signal is not that HTX discloses. It is that disclosure has become cheap. After FTX, PoR transformed from a differentiator into a compliance costume. Every major exchange now publishes something. The question is no longer whether an exchange shows its reserves, but whether those reserves can survive an adversarial reading. HTX's can, barely — and only if you accept its terms.
There is a deeper trap here, one my time in the Patagonian wilderness after the Terra collapse taught me to see. When algorithmic stablecoins failed in 2022, they did not fail because the math was wrong. They failed because the incentives were. A 102% ratio is mathematically solvent and psychologically fragile. It works until enough people believe it does not.
Takeaway
So what does the forty-eighth disclosure actually tell us? Not that HTX is safe. That HTX is still performing the ritual — and that the market, exhausted and compliant, has stopped asking what the ritual proves.
When the herd wakes, the signal has already faded. The thing to watch is not the ratio. It is whether an independent auditor ever signs the page, whether USDT and USDC are ever split apart again, and whether the buffer thickens or thins. Those are the numbers that would tell us something true. Everything else is a photograph of a room the photographer chose.