Murphy

ETH Data 2: Dissecting the Liquidity Pool Structure

On ETH’s URPD chart, the coin stacks in the $2,700–2,800 range are especially high—three stacks together total roughly 13 million (w) ETH, accounting for more than 10% of circulating supply.

Moreover, this batch of coins is sitting at a 40% unrealized loss but has hardly moved.

First, we need to clarify that ETH’s URPD mechanism is based on an account model. Glassnode calculates the weighted average cost using the full balance of each entity.

For example, in February, BitMine held 4.32 million ETH, with an average cost of about $3,100. By August, it had added another 1.48 million ETH, with the buy price roughly in the $1,500–$2,200 range. After the merger, the weighted average cost comes to around $2,700.

The three factors—holding size, cost position, and migration direction—match at the same time.

So the main entity behind this coin stack can basically be considered locked in: BitMine. Of course, it’s possible other clustered entities are mixed in as well.

There are also two additional reasons here:
1) the high-density trading zone from January this year;
2) on-chain staking.

And combined with what we discussed yesterday—Ethereum’s Herfindahl-Hirschman Index reached an all-time high, meaning certain large accounts monopolize supply—this implies that coin concentration is becoming increasingly high.

So it’s very likely related to BitMine, ETFs, and on-chain staking.

The direct benefit this brings is that when the price falls, a large amount of liquidity gets locked and no longer turns into sell pressure.

On the other hand, when ETH’s price returns to that range, whether those coins will still remain firm—whether they will become resistance for an upward trend—depends on the narrative and consensus around ETH at that time.

Every BTC bought in 2025 is currently in a loss. So, as long as the 2025 cohort’s holdings have decreased, aside from wallet transfers, everything else has been sold off to cut losses.

As of today, there are still 4.77 million BTC in the 2025 cohort, down 41.5% from the peak in December last year.

The downtrend slope is clearly divided into two phases: before February it drops rapidly, and after February the decline slows somewhat, but it still maintains a certain slope.

This group should be the largest supply side in the market right now.

If we pull up the data for 2024, 2023, and 2022 for comparison, it’s not hard to see that these remaining holdings that are still in profit have basically already passed the steep part of the declining slope.

And the longer time goes on, the smaller the slope becomes. From the chart, the curve’s slope after February is almost flat like a straight line.

Even if the price drops further, the changes in the number of these holdings are not very noticeable. In other words, everything that needed to turn over has turned over, and the rest just stays put.

From the past two bear markets: at the bear-market bottom in 2022, the 2021 high-level holdings fell by 51%; at the bear-market bottom in 2018, the 2017 high-level holdings fell by 62%.

If we just look at it in a straightforward way, personally I think the bottom of this bear market would be no more than 50–60% (it’s currently 41%), and this doesn’t even account for the BTC bought via the 2025 ETF and by MicroStrategy, most of which is locked and not moving.

New metrics just dropped — the BTC Seller Exhaustion Index!

It measures both low volatility and high losses; when both conditions are met, the indicator triggers a signal.

First, the current state: the seller is already in the "extreme exhaustion zone" (the red area), and this is the first time this bearish cycle has entered that zone.

Looking at historical data, you can see that every previous bear market cycle had similar situations; sometimes more than once (labeled 1/2 in the chart).

When the current 1 appears, it may not be the absolute lowest point of the bear market, but it is definitely within the bottom range.

After that, if the price keeps ranging or goes even lower, but the index does not go any lower, I mark that as 2; across history, the confirmation of 2 is more reliable than 1.

However, the risk is that the price for 2 may still be higher than 1.

Based on the observations above, we can draw this conclusion:

If you’ve already entered a position, that’s not wrong; waiting for 2 to appear before building a position is also not wrong. But if 2 appears and you still don’t buy, then you’ll miss the entire bull market.

(ps: Don’t say I keep coming up with new indicators. It’s the Glassnode team that updates them frequently. I saw them and thought they might be useful, so I’m sharing them with everyone.)

The chip concentration has risen to 14.8%!

With one foot already stepping into the “high-risk zone.” Note! Here, risk does not refer to whether the market is going up or down, but to volatility.

Chip concentration cannot predict direction, but based on historical data, my friends and I have found a certain pattern:

When the curve starts to turn—if before this point BTC’s price was rising—then the probability of continued upward fluctuation is higher; conversely, if it was falling, then the probability of continued downward fluctuation is higher (as shown in the figure).

However, at this moment, the curve is still steadily rising. So, for now, we still can’t predict which direction is more likely to have the higher probability next.

But one thing is certain: risk is accumulating, and volatility is brewing......

Awesome! It’s really over 1 million.

Right now, that single price point of $63,000 has already accumulated to 1.15 million BTC—an extremely rare phenomenon even by historical standards.

Although a recent Coldcard hardware wallet vulnerability has forced some long-term holders to transfer their BTC.

But that’s absolutely not the main reason behind the explosive turnover of supply and demand around the 63k range.

Instead, it’s the result of long-term accumulation while BTC’s price has stayed at low volatility.

At the same time, the concentration of nearby coins has risen to 13.5%.

Coin supply can’t pile up indefinitely. Once the long/short battle reaches the critical point, someone’s going to win.

I’m increasingly looking forward to what happens next.....

As people in the crypto world, we all know that BTC has the best liquidity and the strongest consensus. But the deepest asset consensus hasn’t translated into usage efficiency that matches its scale.

Why is that? Because the BTC mainnet itself doesn’t have a smart-contract environment like Ethereum does.

To make BTC earn yield on-chain or to use it as collateral, you first have to wrap it into WBTC and go through a cross-chain bridge; or else hand the coins over to a custodian. In any case, most holders can’t really accept that.

At least I can’t accept it—Not your keys, not your coins......

So, what @BabylonLabs_io’s Trustless Bitcoin Vaults (TBV, trustless Bitcoin vaults) is here to solve is exactly this problem.

Actually, its mechanism is easy to understand:

1) Keep BTC on the Bitcoin mainnet and lock it in a script that’s jointly signed when the vault is created.
2) Sign the legitimate spend path from the beginning, so nobody can later conjure up a new spending method out of thin air.
3) Then use Ethereum contracts to track this vault, and let the connected DeFi apps treat it as collateral.
4) When redeeming, use existing Script opcodes to verify the redemption proof on the Ethereum side.

That way, “trust” shifts away from whether a custodian is reliable, to cryptography and the two public chains—Bitcoin and Ethereum.

Today, TBV’s first application scenario chooses Aave v4, a mature liquidity market. Looking at the market, this should be the first truly “native + trustless” BTC lending solution.

From then on, BTC holders won’t have to choose only between “keeping holding” and “selling to get liquidity.” This could help change the awkward situation where 99% of BTC is currently still sitting outside DeFi.

That said, I think lending might just be the first step. If in the future stablecoins, credit cards, derivatives, and even insurance can all be built around native BTC collateral, then that’s where the bigger imagination really lies.

#baby $BABY

My precious little cutie, I haven’t looked at the chip structure for a few days, and when I refresh the data, I get quite a shock.

On URPD, there’s a towering candle at the $63,000 level—up to today, it has already accumulated as many as 890,000 BTC.

As far as I remember, in a single price level with such intense long-versus-short battles, this should be the first time since the end of 2025.

If it weren’t for Coinbase locking 550,000 coins in the $83,000–$84,000 range, then $63,000 would probably already be well over 1,000,000 BTC right now.

What does 1,000,000 coins mean? That’s 5% of total circulating supply. Historically, once it reaches a size beyond that, there has basically been at least one major shock.

Because the short-term chips are too concentrated, price sensitivity increases.

At the end of October 2022, right before the FTX collapse, there were already 1,000,000 BTC at around $19,000, and another 870,000 BTC at $18,000. Together, those two levels accounted for 9.7% of total circulating supply.

Then what happened next is something everyone knows—an event as the fuse, layered on top of the fragility of the chip structure, triggered huge volatility.

And now, the $62,000 and $63,000 levels combined have already reached 8%.......

(By the way, today the chip concentration is already 13%, entering the warning zone; it’s only one step away from 15%.)

Come on—give it a decisive one! 🤣🤣🤣

“On-chain data” and “technical indicators” are two completely different dimensions — the former is based on UTXO as its algorithm, while the latter is built on volume and price — yet in many cases, they resonate with each other.

Maybe it’s different paths leading to the same place: when it’s “the right time,” they will, quite unintentionally, point to the same outcome.

For example, in our tweet on July 29, we mentioned that the BTC realized net profit/loss curve hit two extreme negative values in February and June, and at the higher time scale it diverged from price (see the excerpt).

This is a signal of “seller exhaustion, net losses converging” interpreted from on-chain behavior. And at the same time, technical indicators also show a similar kind of guidance.

The CCI indicator used for trend tracking shows a weekly-level oversold signal (red dots). In the past 5 years, this signal has appeared a total of 4 times: November 2018, March 2020, June 2022, and November 2025.

And while an oversold signal only indicates the intensity “at that moment,” what truly deserves attention is the subsequent sustained convergence of the curve and its divergence from price.

In addition, if later on you also see a signal like “downside impulse depletion,” that would indicate that the bearish momentum is being gradually absorbed.

I interpret this as the lead-up process for a trend shift; although slow, the direction is clear.

By combining “on-chain data” with “technical indicators,” the conclusion is:

1⃣ The major-scale downtrend is almost at an end — this can be confirmed.

2⃣ What remains uncertain is only whether there will be another “black swan” like in Nov 2022; but that would just be another confirmation of the ongoing CCI divergence.

3⃣ Betting the whole position on a black swan’s “final drop” is irrational — this isn’t trading, it’s gambling.

4⃣ At this point, there’s no reason to be overly bearish anymore. Time won’t stand on the side of the bears.

In a bull market, the main line is “realizing profits.” Therefore, when the indicators return to break-even (the zero axis), it becomes the support point. When sellers clear the market, it’s easier to form a stage bottom.

In a bear market, the main line is “realizing losses”—the exact opposite of a bull market. When the market reaches break-even, it becomes a pressure point. It’s better to act early; a stage top is easier to form.

And currently, BTC is at the net break-even point.

Based on the logic above, if we only look at smaller timeframes, the probability of “going down” is definitely higher than the probability of “going up” (including after a “fake breakout” followed by a drop).

But if we look from a higher dimension, we can interpret another layer of information:

In the two negative periods in February and June, the pattern is “higher highs, then lower highs.” When prices are lower, net losses are not amplified continuously—creating a divergence with price.

This implies that the preliminary process of a trend reversal is being slowly and quietly incubated.

Even if there is another drop, as long as net losses fall below the previous low again, then on a larger timeframe it can be nearly confirmed that the probability of “going up” will be greater than the probability of “going down.”

Just remember: historically, the end result of sustained divergence is when everything is decided in one decisive move.

This is data that excites all “cycle traders”

As of July 2026, the holdings size of the believers buyers (CB) has reached 4.02 million BTC; this figure is already far beyond the peak of the previous bear market bottom at 3.46 million BTC.

This means that although a large amount of ancient coins within the cycle are waking up and cashing out, more coins are being taken away by the believers buyers as well—especially when prices fall.

Despite the fact that BTC has been criticized by bearish investors, including: the bull-market multiples are not high, the risk-reward ratio isn’t attractive, expectations that it will drop to 40k, 30k, and so on;

none of this can shake the believers buyers’ confidence and pace.

Every time I see the CB holdings hitting a new high, I know we’re one step closer to “spring.”

Big, maybe it’s coming again......

In history, there have been many instances where, after IV falls below 40%, “large fluctuations” in the market followed. The logic behind this was explained in detail in our May 11 tweet (see the excerpt).

👉A simple summary of the 3 points:
1、Low IV means there is a very high consensus on volatility, so upside-down surprise events are more likely to be amplified.
2、Low IV attracts volatility-arbitrage positions; when the short side closes, it can further amplify volatility.
3、As market makers accumulate more short gamma, once a breakout happens, their trend-following hedging can further amplify volatility.

🚩A few cases from the past year:
1、At the start of January, 15 days after IV fell below 40%, BTC dropped from 9.7w to 6.2w;
2、In late April, 14 days after IV fell below 40%, BTC dropped from 8.2w to 6w;
*(i.e., the one I reminded about in my May 11 post)*
3、After June 15, BTC fell from 6.6w to 5.8w;

Of course, after low IV, not every time will necessarily lead to a “downward” type of volatility. For example, in June 2025, 9 days after IV fell below 40%, what followed was an “upward” volatility move ranging from 10.1–11.9w.

So IV is not about predicting “direction,” but predicting “magnitude.”

Now, looking back at the current situation: 1 week is 33%, 1 month is 34%—both are below 40%; so the probability of “volatility” being triggered by market reflexivity and trading rules is rising.

Spot traders don’t really have anything to do with it—futures traders, please fasten your seatbelts!

Judging from Binance’s net transfer data, the current market sentiment is split and disagreement is severe. Two whale cohorts have been operating in opposite directions for a long time.

Group A: individual transfer size greater than 10M, mainly transfers in;
Group B: individual transfer size between 1–10M, mainly transfers out.

These two forces counterbalance each other, resulting in severe disagreement. When Group A is stronger, the BTC price tends to weaken; when Group B is stronger, the price stabilizes (or rebounds).

And currently, Group B holds a slight advantage, so the weak rebound logic can be supported.

At the same time, the BTC balance on exchanges has remained at a relatively high level from 7/13 to 7/15, with no clear decline.

This is completely different from the strong rebound path during 2/25–3/6, when the balance dropped significantly.

From this alone, it’s not hard to see that overall demand is not strong— or that the demand side has not gained a decisive advantage over the supply side.

This situation affects cycle traders not much, but for short-term traders it’s hell-level difficulty: it can easily lead to a situation where both longs and shorts get squeezed, which isn’t suitable for ordinary investors.

PSIP is affected by different cyclical chip characteristics and structure, including the accumulation of low-price chips and an increase in lost chips, causing the lows to be continuously raised.

In this cycle, when BTC fell to $58,000, PSIP had dropped to 46%. This value is infinitely close to the low of the previous cycle.

Based on the current structure, if BTC drops to 4w, PSIP would be: 39%; if BTC drops to 3w, PSIP would be: 36%. This would not only be far below the previous cycle, but even lower than in 2015.

Personally, I think this possibility is very small—very, very small......

From the bear market process in 2022 and 2026, it can be seen that in the initial phase, whale accumulation cannot prevent the price from falling; it can only form interim support, but it cannot change the overall trend.

But as the saying goes, "Strike while the iron is hot—each effort grows weaker, until the last is exhausted." In sustained, multi-phase declines, sentiment and selling pressure are also released and fully discharged in tandem.

After that, if whale accumulation occurs again, its role will be different.

Just like the 1/2/3 I marked in the chart: the effect of 3 is often more effective than 1 and 2. If the earlier selling was like a raging flood, then at this point it is already like a crossbow with its force spent.

Maybe not—but if there are still 4, 5... ahead, then it’s worth paying even more attention. It will be the first ray of light at daybreak!

Judging from the chip (coin) structure, the current situation is very similar to October 2022 (right before the FTX collapse). In both cases, the chips are concentrated within a specific range, and the chip bars in other positions hardly exceed 200k.

For example, currently in the 59k–63k range, there are 1.82 million BTC piled up. In 2022, the situation was even more concentrated: 2.37 million BTC were accumulated in the 18k–20k range.

This indicates that a large amount of active coins has completed turnover here.

The more coins that accumulate, the stronger the support effect; even if the price is briefly broken through in the short term, as long as the concentrated area hasn’t been dispersed, the price can come back not long after—like the pull of a magnet.

Even the “black swan” event back then that was as powerful as FTX did not deeply break through it, precisely because before that, a relatively solid foundation had already been laid in the 18k–20k range.

As for why the accumulated amount this time hasn’t exceeded 2022 levels?

Mainly because liquidity for nearly 5.5 million BTC has been locked up in the 83k–84k range, in Coinbase wallets. So the current 1.82 million isn’t that small anymore.

But the biggest difference is this: in 2022, there was a relatively larger amount of profit-taking positions in the bottom $6,000–$10,000. Now, there are more trapped positions in the $65,000–$92,000 range.

This doesn’t significantly affect the building of the bottom, since those who needed to cut already cut about the same; those with unrealized profits who didn’t sell before now will be even less likely to sell.

However, under this kind of structure, when the market returns to an uptrend in the future, it may also face considerable resistance.

This stablecoin report from Binance Research also validates an earlier idea of mine. Let’s look at the data first:

30% of Binance users allocate more than half of their assets to stablecoins, whereas in 2020 that figure was only 4%. Six years later, this curve has stayed consistently upward, with no correlation to BTC cycles.

Even during bull markets, this set of data doesn’t drop. That suggests that the act of holding stablecoins has moved beyond a “convenient for trading” transitional attribute, and instead has become a “destination.”

Some users buy stablecoins on Binance using their local fiat currency, but they face severe premia. Especially in countries with malicious inflation, the average premium reaches 62%. In other words, they clearly know that, at the official exchange rate, they are paying an extra 60%—yet they still choose to exchange.

No one would pay a 62% cost just for “convenience in trading.” What they’re paying for is the ticket to exit their local fiat system; what’s being priced in is the expectation of local currency depreciation and capital controls tightening.

So, for a significant portion of users worldwide, stablecoins serve as savings accounts.

I once proposed an idea: don’t assume that when an exchange’s stablecoin balance is rising, it automatically means funds are coming in and preparing to buy the dip.

As balances increase, the motivation may also have nothing to do with market conditions.

For example, as mentioned above, unstable local currency and financial exclusion drive this demand for saving. The purpose of money entering is simply to “sit there,” not to wait for an entry opportunity—so it doesn’t constitute potential buy-side demand.

But conversely, when balances fall, it’s more straightforward.

Because savings-style capital is highly sticky, there’s little reason for this money to exit in bulk. So when balances show a clear decline, it is likely that another portion of funds—namely the active capital actually participating in trading—is moving to other markets.

For instance, not long ago, when US stocks “drained” capital, the drop in exchange stablecoin balances corresponded to the migration of this type of funds.

In short, stablecoin “savings-ification” makes this indicator effectively one-directional: when it rises, the signal is diluted (unless the rising curve suddenly becomes much steeper); when it falls, that’s when we should truly be on alert.