avatar
YJ | OpenLayer
3,120 Twitter followers
Follow
Founder @OpenLayerHQ | Ex-@RobinhoodApp
Posts
avatar
YJ | OpenLayer
The story of #STO teaches us that if you don't like Binance, you shouldn't criticize it; you should stop making trades with clearly negative EV, which is more useful than criticizing Binance. Why short with negative EV? You're short with a funding rate of -1000% APR. You don't know how high the market manipulators can push the price. Exchanges take a cut of the transaction fees, and market manipulators can extract tens of millions of dollars in liquidity from a single pump. If you invest 1 USDT, you might only get less than 0.8 USDT in EV return. After a few more trades, it will go to zero. The market maker controls the distribution of shares and the rhythm of price manipulation. Unless you can accurately predict the market maker's intentions, a double blow to both long and short positions is highly probable. Game theory is interesting, but why play a game with information asymmetry? If you truly dislike exchanges and market makers, you should leave the altcoins game. Complaining is useless; quitting is the only effective solution. If no one plays, naturally exchanges and market makers won't make money.
STO
4.19%
avatar
YJ | OpenLayer
03-31
1. Low ego, high agency, natural curiosity, strong execution 2. Forward-thinking, avoid path dependence, find new structural changes 3. The vintage year for crypto is the first two cycles, which have already passed. 4. Losses in 2025 came from DAT, overestimating Wall Street's interest in long-tail crypto assets. 5. Currently favored structural changes in crypto: PayFi and CeDeFi 6. The combination of AI and crypto is being adopted by more companies as stablecoins become programmable money.
YettaS
@YettaSing
03-30
跟Mercy聊了非常开心的一小时！ 做妈妈和做投资人，本质上都在做同一件事：在极度不确定中做长期判断，并为结果负责。 在视频中，我复盘了这个周期的得与失，以及接下来 Crypto 可能的路径。 很多时候，市场教你的和生活教你的，是同一课。 x.com/Mercy_okx/stat…
EGO
14.32%
avatar
YJ | OpenLayer
03-27
Today, Unity released its earnings guidance seven weeks before its official Q1 financial report. The data is quite good: Q1 revenue of $505-508 million, more than $20 million higher than the Q4 guidance, and EBITDA of $130-135 million, directly exceeding the guidance by 25%. It's worth noting that IronSource, acquired in 2022, and its subsidiary Supersonic have been dragging down performance. Previously, Unity stated it would gradually divest its IronSource business, and today it announced that it will shut down the IronSource advertising network and sell Supersonic on April 30th. Cutting the burden and going all in on Vector. Vector (the core of Strategic Grow), Unity's self-developed product as an AI advertising platform, representing its transformation from a traditional software company, saw a 48% year-over-year growth. Management's confidence in Vector is also the reason for the aggressive closure of IronSource. Unity is essentially an AI advertising company that provides development software to game developers to obtain high-quality data. If you find companies among all the software companies that were unfairly punished in February that might benefit from AI rather than be disrupted, then even though Unity rose 12% to $19 after hours, the risk-reward ratio is still quite good.
VEC
0%
avatar
YJ | OpenLayer
03-25
Pure model companies are unlikely to produce ten-bagger stocks again, as commoditization is an inevitable trend. Good ten-bagger stocks all possess strong network effects and monopolies. However, the products of large model companies are converging. The US has Anthropic, OpenAI, Gemini, and Grok, and China has several; globally, about ten companies are rapidly catching up in terms of quality. Demand is indeed exploding, but competition on the supply side is equally fierce. The model itself is becoming a public good. To use a perhaps imperfect analogy, Ethereum (ETH) had virtually no competitors in 2017; all ICOs ran on it. Now, countless public chains can do the same thing, and even with its narratives of decentralization and security, ETH has lost its absolute appeal. In contrast, giants like Nvidia, Meta, and Google each have absolute monopoly power in their respective markets, exhibiting a winner-take-most structure. Large model companies currently do not show this structural lock-in effect. Unless a company can build an irreplaceable application layer or data flywheel on top of the model, transforming itself from a model provider into an operating system and gateway for the AI era, it will be unlikely to capture a significant market share. Currently, I think Anthropic is the most likely candidate. Its annualized revenue has surpassed OpenAI, exceeding B2B, and at GTC, Nvidia CEO Jensen Huang mentioned that everyone at Nvidia is using Claude. If Claude truly integrates into enterprise workflows and becomes an operating system-level entry point, network effects could emerge. Large-scale model valuations are already quite high, but it's still worth watching whether any company is building a moat beyond the model itself. One question I haven't fully grasped is whether the winners of network effects in the AI era are model vendors or application-layer companies that utilize large models. Theoretically, Claude targets the entire white-collar market's TAM (Total Asset Management), with extremely high growth potential. However, in niche sectors like healthcare and law, application-layer companies with industry know-how might leverage richer vertical data using RAG (Rapid Application Model) + smaller models to achieve less illusion and, coupled with stronger sales and distribution channels, gain more market share. User loyalty is for the product, not the model, unless the model extends its capabilities to become a product itself.
ANTHROPIC
4.27%
avatar
YJ | OpenLayer
03-20
$VCX Listed on the NYSE today, up 143% on its first day. This is a public VC fund that allows retail investors to buy a basket of top-tier unlisted AI companies on the open market. Holdings distribution: Anthropic 21% Databricks 18% OpenAI 10% Anduril 7% SpaceX 5% The top 7 holdings account for 75%, essentially a basket of AI pre-IPO companies. Why the surge? Because retail investor demand for these companies is extremely high, and there are almost no conventional channels to buy them. $VCX essentially opened a floodgate, causing an instant imbalance between supply and demand. The fund's NAV was approximately $25 per share, now at $76. Retail investors are buying with 3x NAV, indicating extremely high FOMO (Fear of Missing Out). The risk-reward ratio of chasing the first-day gains of a closed-end fund with 3x NAV is very poor. Closed-end funds mostly revert to trading at a discount in the long term. However, as long as OpenAI/Anthropic has not yet IPOed, this scarcity premium will not completely disappear in the short term. What's worth noting is Robinhood's similar fund, $RVI, which fell 16% on its first day of trading two weeks ago and is currently trading at a discount of about 7%. The reason is that its holdings don't yet include popular AI stocks like O/A/S. If it's bullish on the market, these will likely be added later, making now a good time to buy.
NAV
0.13%
avatar
YJ | OpenLayer
03-17
I have a vague feeling that something big is coming. Last week, MSTR issued $1.18 billion worth of STRC, bringing its total size to over $5 billion. My understanding of MSTR is that by selling fixed-income products like STRC (similar to US Treasury bonds) to relevant financial groups, it shifts BTC volatility from high-frequency, short-cycle trading to low-frequency, long-cycle trading. In the past, BTC cycles were mainly driven by three forces: the supply shock from halvings, retail investor FOMO (fear of missing out), and the spiral of leveraged liquidation during bear markets. STRC introduces a new, continuous source of demand: traditional fixed-income funds. These funds are characterized by their massive size, stable behavior, and stickiness. This means that BTC's demand curve has changed from a pulse-like pattern to a continuous injection. This capital has the potential to initiate a slower and longer bull market for BTC compared to previous cycles. However, at the same time, I believe MSTR will inevitably crash eventually. One of the fundamental ironclad rules of finance: historically, there has never been an asset that only rises and never falls. If an asset truly only rises and never falls, everyone will buy it. Everyone buys, pushing the price to an instant infinity, and then it stops rising. This is a self-contradictory logic. Even an index like the S&P 500 has existed for less than 100 years, and it represents the power of the United States during the peak of the strongest technological revolution in human history, the benefits of globalization, and the hegemony of the US dollar. Even so, buying the S&P 500 in 2000 wouldn't break even until 2013. Meanwhile, Bitcoin, hailed as the "atomic world" and the most widely accepted store of value in 5000 years of human history—gold—would only break even 28 years later in 2008 if bought in 1980. Once BTC enters a long-term downward cycle, MSTR will inevitably be forced to sell Bitcoin, while simultaneously being unable to maintain interest payments on STRC. Just as we cannot predict death, we cannot predict when this will happen, but currently, the upward cycle of the next four years is already being driven by these buyers seeking fixed returns. All we need to do is seize the opportunity before the collapse and make the music as loud as possible.
BTC
0.4%
avatar
YJ | OpenLayer
03-08
Here's a scary story: Walmart's WMT forward PE ratio is twice that of Nvidia's NVDA. WMT's market capitalization is approaching one trillion dollars, ranking among the top ten US stocks, with a 3-fold increase in 5 years, an annualized return of 25%. Costco next door also tripled in 5 years. These two grocery stores have returns that far surpass most tech stocks. Currently, WMT's forward PE ratio is 42, while NVDA, the world's hottest AI chip company, is only 21. A grocery store is twice as expensive as a GPU seller. The market is worried about the cyclical nature of AI capital expenditure, giving it a cyclical discount, while favoring supermarket stocks—where people need to eat regardless of economic conditions—due to predictable cash flow certainty. Whether AI is in a bubble is a frequently discussed issue. The explosive productivity and transformative impact of AI on white-collar jobs are undeniable. The current lack of consensus is whether the short-to-medium term AI capital expenditure by major companies is excessive and constitutes a bubble. If you firmly believe that future token consumption will be ten to a hundred times higher than it is now, then NVDA still seems reasonably priced.
PE
0%
avatar
YJ | OpenLayer
02-26
CRCL's financial report is out, and most data chains can monitor it in real time. However, one figure deserves special mention: non-reserve interest income: 24Q4: $3M 25Q3: $28M 25Q4: $37M 2026 guidance: $150-170M One of the core reasons the market was bearish on CRCL before was its over-reliance on interest income, which would be unfavorable during a rate-cutting cycle. Now, although the guidance for non-interest income (partner services + transaction revenue) only accounts for about 5% of total annual revenue, the growth curve is already very clear. Trends are more important than absolute values, which is why CRCL's profit margin slightly increased to 40% quarter-on-quarter. Another catalyst is the growing narrative surrounding AI agent payments. The idea that stablecoins are naturally suitable for inter-agent payments has been around for almost two years, and the explosive popularity of OpenClaw has made the market take it seriously. CRCL has thus become the most direct target in this field. Looking at today's USDC supply: the bad news is that it hasn't increased for six months, remaining around 75B. The good news is that even though BTC has fallen to 60K, the USDC supply has not decreased, which to some extent indicates that non-crypto trading use cases are gradually supporting the basic market.
CRCL
258.31%
loading indicator
Loading..