I read Ran's portfolio changes as preparation for a possible pullback, rather than a retreat from crypto. In his Crypto Banter update, he cuts his Telegram-related holding, sells PUMP and renews his interest in Sui. Most of his portfolio stays invested, with a small cash reserve and a price-specific buy list. His plan depends on a correction that he expects, not one that is certain to happen.
Why the creator expects a correction
Ran believes crypto is in a bull market, but thinks a move down is more likely than a move up in the near term. A correction is a decline within a broader rising market. Investors cannot know ahead of time whether a fall will stay within that pattern or become a longer downturn.
He bases his bull-market view partly on how tokens respond to news. Starknet's token price doubled after its announcement that it was considering becoming a layer-1, an independent base blockchain. He also points to an Aerodrome price rise after an announcement about growing tokenized-stock trading activity. These products use blockchain tokens to provide exposure to shares, although the rights attached to them depend on the product.
I see those reactions as signs of interest in the projects, not a reliable test of the whole market. News can move individual tokens under different market conditions. His argument depends on those examples reflecting wider demand rather than enthusiasm for a few projects.
He also interprets Ethereum's response to Tom Lee's comments about BitMine as resilience to bad news. Lee described stopping purchases at a 5% supply cap, while saying another 100,000 ETH was needed to reach it. That makes this an announced stopping point, rather than a statement that purchases had already ended. Ran describes ETH falling to around 2,400, perhaps 2,350, before returning to 2,500.
The opening TradingView screen puts Bitcoin at $82,957. Ran gives its recent low as about $80,000 and believes a further correction is still ahead. I keep those prices separate from his forecast: they record where Bitcoin traded, while neither the decline nor a rebound establishes its next move.
Cutting the Telegram-related position
Ran substantially reduces the holding he calls Telegram and moves the proceeds into cash. He initially describes removing it, then clarifies that a small position remains so he retains some exposure if it rises.
His original thesis depended on distribution. Telegram's audience could offer a large route into crypto if Pavel focused on bringing crypto activity to those users. Ran believes that emphasis has taken too long to deliver what he wanted. He still thinks it could happen eventually, but no longer wants to keep as much capital waiting for it.
I find the timing question more useful than a verdict on whether the idea is good or bad. An investor can believe in a project's long-term prospects while deciding its execution no longer justifies the current holding. That is an opportunity-cost judgement: money committed to one idea is unavailable for another.
Reducing the holding introduces a different risk. The delayed development might arrive after the sale, leaving the smaller position to capture only part of a subsequent rise. Keeping a little mainly to avoid feeling left out can also preserve an emotional attachment to an idea that no longer meets the investor's criteria.
Distribution and execution need separate scrutiny. Access to a large audience does not automatically produce users for a crypto product, and adoption does not automatically benefit a particular token. I read Ran's decision as disappointment with progress, not a claim that Telegram-related crypto activity has no future.
Why PUMP revenue did not change his decision
Ran sells PUMP, the token associated with Pump.fun, after entering at 2.5 and selling at 5.6. He describes that as a 2x. Taking a profit explains part of the decision, but he also questions the business behind the token.
A launchpad helps people create and introduce tokens. In his assessment, Pump.fun's founders built a successful launchpad but have not established other successful products. Harmful trading activity makes him doubt how durable its appeal will be. He distinguishes the platform itself from the people carrying out what he calls "extraction schemes" on it.
The comparison he discusses puts PUMP's market capitalization at $2.9B and HYPE's at $22B. Market capitalization measures the market value of circulating tokens, not cash available to holders. He cites Pump revenue 32% above Hyperliquid's during the week being compared, despite the lower valuation.
I would not infer that either token is cheap from that comparison. Revenue does not establish how much value reaches holders, whether the activity will repeat or how much risk the price reflects. A selected week's revenue also gives only a narrow view of two different businesses.
Ran's explanation rests on network effects: a service becomes more useful as participation grows. He argues that more Hyperliquid users can bring liquidity, allowing trades without sharply moving prices, and that this can attract further users. He believes Pump.fun lacks comparable benefits and can instead expose more participants to damaging token launches.
The claim that a launchpad has no network effects remains his opinion, not a demonstrated rule. I would examine whether new participants improve other users' experience and whether that benefit lasts. His preference for social trading models has a related limit: he acknowledges that traders can have private wallets alongside public ones. Visible holdings therefore need not reveal their full exposure or intentions.
Sui's renewed agent thesis
Ran renews his commitment to Sui after putting the holding on probation. His change of view follows Basecamp, Sui's developer event, and focuses on AI agents. These are software systems that can act toward a goal, including interacting with on-chain applications.
He says Sui's work increasingly targets those agents. He cites a live on-stage demonstration reaching 40 million transactions per second, audited by CertiK. He also describes work with Google Cloud intended to keep agents within their limits and an on-chain verification layer intended to identify anomalous activity.
I separate that technical claim from the investment case. A demonstration can establish capability under its test conditions, but everyday capacity, demand and token value require different evidence. The video does not establish how those conditions compare with ordinary network use.
His security description also has limits. Identifying unusual activity may help users notice a problem; it does not necessarily prevent a malicious transaction, reverse it or recover assets. Detection should not be read as protection against every kind of compromise.
Ran accepts that Sui arrived late in the competition for human users and lacks a defining application. He argues it may be better placed to compete for agents, partly because its object-based design organizes assets and other state as objects. The claim that this design makes Sui especially suitable for agents is part of his thesis.
I welcome the willingness to reconsider a holding after new information. But this case still depends on agents needing the capabilities Sui is building and their activity creating meaningful token demand. Technical progress and commercial adoption are separate tests; neither guarantees a favorable investment outcome.
What the cash allocation changes
Ran says 93 to 95% of his portfolio remains invested, with 5 to 7% in cash. He briefly gives other figures, then corrects himself and repeats this range.
The reserve gives him money to deploy without selling another holding first. It leaves the portfolio heavily exposed to crypto, however, so falling prices would still affect the invested portion. I see greater flexibility here, not an exit from market risk.
Some cash comes from trimming profitable positions. His examples include taking money off Ethena after a 3x and reducing other holdings after substantial gains. That converts part of an unrealized gain into funds available for another purchase, rather than leaving every winning position untouched.
The trade-off is straightforward. Lower prices could give the reserve a use; a rise without the expected pullback would leave that money out of the gains. Ran accepts this because he considers near-term downside more likely. The outcome depends on both the market and what he does when prices move.
I also distinguish cash from diversification. Several tokens need not behave independently in a broad sell-off. Holdings based on similar trading or adoption stories can all suffer when investor appetite changes.
He has kept most of his exposure rather than turning the entire portfolio into a bet on falling prices. That limits the effect of being wrong about a correction, while the smaller reserve leaves room for the purchases he is considering.
Crypto portfolio buy levels and their limits
Ran's purchase list is conditional. He does not need every candidate and says he would be satisfied if three or four of his 10 options reached acceptable prices. The thresholds state what he is willing to pay, not where he predicts each token will trade.
| Asset | Creator's stated threshold |
|---|---|
| Backpack | Under 0.8 |
| Uniswap | Under $5 |
| Venice | Under 15 |
| Grass | Under 0.5 |
| Geodnet | Under 0.2 |
| Derive | Under 0.2 |
Derive is already a holding he wants to increase, although he doubts it will return to his desired level. None of these thresholds is an independently established estimate of fair value or a support level, where traders expect buying interest to appear.
I find the advance planning useful because a written price can reduce the urge to chase a rising token or improvise during a fall. Reaching that price still calls for another look at the investment case. The number alone cannot settle the decision.
His wider list includes LayerZero and a tentative interest in Stonk. He first describes LayerZero as a watchlist item, then clarifies that he has bought a little and wants more exposure. With Stonk, he remains unsure what it will become beyond a launchpad, the same concern behind his PUMP sale.
Watching an asset is not a commitment to buy it. Existing exposure matters too: Ran says he already has enough NEAR for his purposes. I would not treat his list as a stand-alone portfolio, because it reflects what he owns already.
A lower price can offer a better entry or signal a weaker proposition. Broad selling may leave a project's case intact, while project-specific damage may undermine it. A threshold is useful as a point for reassessment, rather than an instruction that overrides new information.
Security warnings need different kinds of checks
Justin Drake's call to plan for a possible bunker mode is one of the security concerns behind Ran's market discussion. Drake proposes a controlled migration toward fresh addresses whose public keys remain hidden behind a hash. His message urges calm planning rather than a rush to move assets.
The warning concerns the assumptions that protect private keys, the secrets that authorize control of funds. A related public key can normally be exposed without revealing that secret. A hash, a one-way mathematical transformation used throughout crypto, matters here because it can keep the public key hidden.
Ran interprets Drake's message as a concern that sufficiently powerful AI-assisted mathematics could threaten elliptic-curve cryptography before quantum computers do. He calls it a real threat, but not an immediate one. I would not turn that interpretation into a claim that AI can currently derive crypto private keys from public keys.
An emerging cryptographic concern is different from an active wallet compromise. Hasty migration can introduce familiar mistakes, such as sending funds to the wrong address or following a fraudulent migration instruction. Technically specific, authoritative guidance matters more than urgency around a broad warning about future computing capabilities.
Ran also reports tampered Ledger devices obtained through an Asian distributor. He says about $90 million has been drained from affected wallets, and a post by Mark Karpeles shows a modified device board. This is his reported account, not an independently verified finding about all devices or distribution channels.
I give the ordinary handling risks equal attention. Trusted purchasing routes and the manufacturer's setup checks matter, and a device supplied with a ready-made recovery phrase is suspect. That phrase can restore control of the wallet, so someone who already knows it can control the funds.
Paper Trade rewards do not cancel trading losses
Ran describes Paper Trade as a perpetual-futures exchange built on Hyperliquid. These contracts give traders price exposure without a fixed expiry. Leverage increases the position controlled by margin and magnifies gains and losses. If the margin becomes insufficient, the position can be liquidated.
He describes leverage of 100x and 1,000x and token rewards tied to losses through a pool mechanism. In his account, losses fill the pool. Rewards stop when it is full and become available again when it needs funds. He warns against deliberately losing money to obtain tokens.
I share that caution because the loss is immediate while the reward's availability and value may be uncertain. Receiving tokens does not establish that a participant is better off. That depends on the reward rules and the tokens' eventual usable value.
He says larger traders are farming rewards through opposing positions across venues. Taking a long position on one and a short position on another can make the combined exposure delta neutral, or less sensitive to the underlying asset's direction. His account does not establish that this strategy is profitable or risk-free.
Costs and mismatched execution remain, as does liquidation risk. At extreme leverage, one side may be lost before the other delivers a realizable benefit. I would not mistake a neutral-looking combined position for a protected account.
The dashboard reaches $206 billion in total traded volume. That measures turnover, not deposits or profits. Repeated leveraged trades can generate large volume without equivalent new capital entering the system. Ran sees the activity as a notable example of applications built on Hyperliquid; the figure itself does not validate the reward strategy.
Turn a watchlist into a decision record
Ran keeps prospective purchases at the top of his TradingView list so he can focus on selected candidates when prices fall. He thinks a separate watchlist would be better organized. Either arrangement puts the relevant assets within reach before volatility adds pressure to a decision.
I find a short decision record more useful than a price alert alone. It can hold the reason for considering an asset, what would weaken that reason and the maximum exposure under consideration. Those details give the alert context instead of letting a price become the entire argument.
The record also helps distinguish evidence from expectations. A completed sale records an action. A displayed price records a market observation. An announced plan records an intention, while a forecast about direction remains uncertain regardless of the speaker's enthusiasm.
For me, the test is whether the plan allows for being wrong. Prices might keep rising without the desired entry, or a decline might last longer than expected. Even when a watched token reaches its threshold, project-specific damage may have weakened the reason for buying it.
