The ICO Game: Buying Early, Selling Smart — Remember the EOS Arbitrage Era?

mediumPublicado a 2025-05-30Actualizado a 2025-05-30

Resumen

It was 2017, when the controversy over BTC expansion was the most heated. I was a die-hard expansionist at the time, and I liked to write articles to express my support for expansion.

Written by: Huang Shiliang

EOS changed its name to A, the A in abcd. In the future, EOS will be ranked first in the alphabetical order in the exchange.

I have always had a good impression of EOS because in the first three months of its 350-day ultra-long ICO fundraising process, I participated in arbitrage in the primary market of EOS's ICO and the secondary market of the exchange, making a small profit.

That was the easiest money I have ever made.

EOS started its ICO on June 26, 2017 and ended on June 1, 2018. It raised more than 4 billion US dollars in ETH and issued 900 million EOS tokens. It was really an awesome fundraising, unprecedented and definitely unparalleled.

The ICO mechanism of EOS was still very novel at the time.

EOS initially conducted an ICO through a smart contract deployed on Ethereum, issuing tokens in two phases: the first phase was a one-time distribution of 200 million EOS in the first five days of the ICO (June 26-30, 2017);

Then entered the second phase, which is a daily round issuance mechanism. Every day (about every 23 hours), the EOS smart contract accepts users' ETH inputs and distributes 2 million EOS on that day in proportion to the ETH share of all participants after the round ends.

This daily issuance phase lasted for 350 days, with a total of 700 million EOS issued. Together with the 200 million in the first five days, a total of 900 million, another 100 million EOS are reserved for the project party Block.one, forming a total issuance of 1 billion.

EOS's ICO sold 900 million EOS-erc20 tokens and received $4.2 billion, a total of 7.2 million ETH coins.

At that time, not many people in the industry studied how to operate Ethereum contracts, how to recharge contracts, and how to claim purchased tokens.

The tools at that time were very simple, unlike the current contracts and wallets, which have very good UI interfaces.

In 2017, the main wallet of Ethereum was MyEtherWallet. I had a good impression of this wallet. The private key file used was still a json file. To operate the contract for investing in EOS ICO, you need to configure various parameters yourself, which is better than command line (DoS) operation.

The current metamask wallet, imtoken, etc., have been used with various contracts, and the various operation commands have been very well visualized in the UI. Users only need to click a button to complete all contract interaction actions.

At that time, MyEtherWallet invested in EOS ICO, and each time you had to import the private key json file, select the contract, select the contract parameters and commands, and fill in the amount of ETH you want to inject into the contract, including the amount of gas, which is user-defined.

Then when it comes to claiming (claim eos-erc20), you need to fill in several parameters again.

Only these operations, few people did it at that time.

I was curious at first. I was very curious about EOS because I had played various bitshares (bitshares, the first project of EOS developer BM) earlier (around 2015). So I carefully studied the ICO of EOS on ETH.

During the ultra-long ICO period of 350 days, EOS can also be traded directly on the exchange. That is, during the ICO period of EOS, the primary market and the secondary market were opened at the same time. But the prices of the primary market and the secondary market are not synchronized.

This creates a potential arbitrage space.

At that time, I recharged ETH into the ETH ICO contract every day, and when the time came, I went to claim eos-erc20, and then recharged it directly to the exchange to sell it.

It was just such a simple arbitrage, without any hedging, no j8 strategy, just a simple bet that the price of the primary market was cheaper than the secondary market, and it could make money all the time.

This kind of money-making lasted for a very long time. In the first three months, I almost never lost money, but in the next few months, I started to lose money occasionally, and then I stopped doing it half a year later.

Why did I dare to judge that the primary market price of eos would be cheaper at the beginning? It was because not many people would buy coins in the ICO contract.

At the beginning, checking the interactive addresses on the chain, there were less than 100 addresses per day. In the first three months, most of the time it was 148 addresses. I remember this number. Later, the addresses began to increase, and then the profit was gone.

In this primary and secondary betting price difference arbitrage game, I also observed the behavior of hackers attacking the chain on the chain. It was really for the sake of making money.

Because there is a mechanism that ends a phase every 23 hours, in the last few minutes of a phase, the price in the primary market can often be calculated, because the amount of ETH in the contract is transparent, and then compared with the price in the secondary market, it can be known whether there is arbitrage space.

So often in the last few minutes, a large amount of ETH will suddenly be recharged, which may cause the primary price to be more expensive.

Then every last few minutes, a hacker will construct a terrifying transaction that consumes ETH gas, blocking the entire ETH chain, so that others cannot recharge ETH into the eos ico contract.

After a failed recharge, I learned my lesson. Every time I grab a share of the first level at the last minute, I directly pull up the gas, which is a waste of money. After losing gas several times, I stopped the game and let them compete. I tried my best.

That six-month arbitrage war did make a small profit, but blessings and disasters are interdependent. Many things, after reflecting on them in the next few years, I found that they left disasters.

Because EOS's ICO has a 23-hour period, the claim time changes every day in 1 hour, so the time to grab the first level and claim the token changes every day, so 1/3 of the period occurs when I should be sleeping.

But in order to make money, I don't care about sleeping, I get up at 12 o'clock in the evening and 3 o'clock in the morning.

This thing, staying up late once or twice is really okay, but if it can last for a week, it will be troublesome. And this is not a simple stay-up, because there is a huge profit, every time I do it, I am either very excited or furious...

Of course, it's not just my health that is bad, I think there will be worse effects later.

I was leading a small team to work on a project at the time. But my brothers saw that I was like this every day, and the boss was not motivated, so it was even more impossible for others to work hard.

Looking back, there was still a chance to make this project, and I was messing around and hurt my brothers. Later in 2018, several brothers took the initiative to leave.

Come to think of it, I was actually quite fragile, which made me no longer have the courage to lead a team to work.

Just this year, Musk entered politics in a high-profile manner, and I thought that his companies, including Tesla, X.com, and SpaceX, might suffer, which is a loss for the whole world. If the boss does not personally lead the charge, the company will most likely not do well.

This experience also had another byproduct.

It was 2017, when the controversy over BTC expansion was the most heated. I was a die-hard expansionist at the time, and I liked to write articles to express my support for expansion.

Maybe because I made money from EOS arbitrage, I became inflated at the time, and because I couldn't sleep well every day, I wrote articles and expressed my opinions in the group, which might be too arrogant and offended a lot of people.

Later, I was cyberbullied. The experience of being cyberbullied made me very cowardly on the Internet.

Until now, I feel that I am quite cowardly on the Internet and dare not curse people.

There were two experiences at that time that left a deep impression on me.

One time, the RSK (Bitcoin's sidechain project) team was coming to China for a roadshow. Because I wrote a lot of articles about sidechains in the early years, their team invited me to follow them for a speech. I criticized them on the phone very rudely for not being firm on capacity expansion. Oh, I was so arrogant and stupid.

The night before the call, at 3 a.m., I was still working on the EOS ICO. I was so excited that I didn't sleep. How could I have the energy to speak properly when I answered the phone.

Another time, a reporter from a newspaper called me to interview me and asked me about capacity expansion. I said in a thunderous voice, "Do you dare to publish my original words, because I said a lot of things that are not good for small block supporters?" Oh, in fact, if I only talk about technology, it might be better.

I was too arrogant at that time. In fact, I have always been a very humble person.

Time flies, and EOS ICO has been over for 8 years. The money earned has been lost, leaving only some memories.

Lecturas Relacionadas

¿Dónde está el anclaje de valoración de MegaETH tras la salida de Aave y la fuerte volatilidad de su TVL?

Según datos de DefiLlama, MegaETH experimentó una fuerte volatilidad en su TVL entre el 9 y el 10 de julio, cayendo brevemente a poco más de 30 millones de dólares, lo que supone una caída del 60% en 24 horas y un descenso del 70% respecto al pico de mayo. El principal protocolo en cadena, Aave V3, retiró alrededor del 80% de su liquidez en un día. El precio del token MEGA cayó a unos 0,048 dólares, con una capitalización de mercado de unos 54 millones de dólares y un FDV de aproximadamente 4.800 millones. MegaETH, una vez una de las nuevas blockchains más esperadas, contó con una fuerte afluencia de capital inicial respaldado por importantes fondos de capital de riesgo y entusiasmo de KOLs. En su punto máximo, su FDV alcanzó los 20.000 millones de dólares y su TVL de DeFi llegó a 245 millones, posicionándose entre las 11 principales blockchains. Sin embargo, su TVL dependía en gran medida de Aave y de estrategias de apalancamiento con stablecoins como USDe. Cuando el rendimiento de USDe cayó por debajo del coste de préstamo en Aave, estas posiciones se desmontaron, provocando una salida masiva de capital. Actualmente, el valor de MegaETH parece desalineado con su actividad real. Con un FDV de 4.700 millones de dólares, los protocolos de la cadena generan menos de 90.000 dólares de ingresos reales en 30 días y tienen solo 2.619 direcciones activas diarias. Además, gran parte de los ingresos provienen de una aplicación de juego de cartas, Monster, y no de protocolos DeFi. Su stablecoin nativa, USDM, también está perdiendo valor. La comunidad critica la falta de comunicación del equipo y la migración de proyectos a otras cadenas. El caso de MegaETH refleja una tendencia más amplia en la que el mercado está dejando de premiar los altos TVL impulsados por incentivos o arbitraje, exigiendo en su lugar fundamentos sólidos, ingresos reales y un ecosistema saludable para respaldar la valoración. A menos que el equipo pueda convertir el capital recaudado en adopción y utilidad real, es difícil ver un pilar sólido para su valoración más allá de posibles rebotes temporales por sentimiento del mercado.

链捕手Hace 4 hora(s)

¿Dónde está el anclaje de valoración de MegaETH tras la salida de Aave y la fuerte volatilidad de su TVL?

链捕手Hace 4 hora(s)

Informe en profundidad de Goldman Sachs: ¿Quiénes serán los ganadores a largo plazo en la industria china de modelos de IA a gran escala?

Informe de Goldman Sachs sobre los posibles ganadores a largo plazo en la industria china de modelos de IA. Los modelos de inteligencia artificial de China se encuentran en un punto de inflexión histórico. Según Goldman Sachs, los modelos de código abierto han alcanzado un rendimiento cercano al de los modelos propietarios globales líderes. Esta evolución ha pasado del enfoque en la eficiencia de costos de DeepSeek el año pasado al momento de la inteligencia del modelo de GLM de Zhipu este año. La adopción por parte de empresas chinas y pymes globales se está expandiendo rápidamente, creando un efecto de vuelo de datos que impulsará nuevas iteraciones. La clave del éxito radica en la innovación arquitectónica y la eficiencia de parámetros, lo que permite a los modelos chinos lograr un rendimiento comparable a un costo significativamente menor. Se está formando una estructura de mercado de dos niveles: un segmento premium (ej. GLM5.2 de Zhipu) y un segmento de bajo costo para tareas de agentes inteligentes y usuarios sensibles al precio. Se proyecta que los ingresos por API y suscripciones de modelos chinos crecerán drásticamente, de 35 mil millones de RMB en 2026 a 879 mil millones en 2030. La estrategia predominante es el código abierto o de pesos abiertos, que favorece la adopción, aunque presenta desafíos para la monetización. La industria podría migrar hacia un modelo de "pesos abiertos + licencia comunitaria" con acuerdos de reparto de ingresos. La expansión internacional, especialmente en mercados no estadounidenses, representa la mayor oportunidad de crecimiento, aprovechando la ventaja de costo-rendimiento. Goldman Sachs identifica a los posibles ganadores utilizando un marco tridimensional que evalúa la capacidad de fijación de precios, la ventaja de costos y la solidez financiera. En modelos de texto básico, Zhipu AI y DeepSeek tienen la posición más sólida. En el ámbito multimodal/generación de video, ByteDance lidera con su modelo Seed. La firma mantiene su calificación de compra para MiniMax, citando su atractiva valoración.

marsbitHace 4 hora(s)

Informe en profundidad de Goldman Sachs: ¿Quiénes serán los ganadores a largo plazo en la industria china de modelos de IA a gran escala?

marsbitHace 4 hora(s)

Informe Profundo de Goldman Sachs: ¿Quién será el ganador a largo plazo en la industria china de modelos de IA grandes?

China se encuentra en un punto de inflexión histórico en cuanto a modelos de IA a gran escala. Según Goldman Sachs, los modelos chinos de código abierto/ponderaciones abiertas han alcanzado un rendimiento cercano al de los mejores modelos propietarios globales, con una rápida adopción por empresas chinas y pymes mundiales, creando un efecto de círculo virtuoso de datos para futuras mejoras. El informe destaca la ventaja de coste-eficiencia de los modelos chinos, lograda mediante innovaciones arquitectónicas (como MoE) que permiten alto rendimiento con menos parámetros activos (3-5%). Modelos como DeepSeek V4 Pro (1,6 billones de parámetros) y GLM5.2 de Zhipu (0,7 billones) son ejemplos clave. El LongCat 2.0 de Meituan, entrenado con chips nacionales, marca un hito en la autosuficiencia. El mercado se está estructurando en dos niveles: modelos premium (como GLM5.2, ~1$/millón de tokens) y modelos económicos para agentes de IA (~0.06-0.2$/millón). Se proyecta que los ingresos por API/suscripción crezcan desde 35.000 millones de RMB en 2026 hasta 879.000 millones en 2030. La estrategia predominante es el código abierto o ponderaciones abiertas (salvo ByteDance), lo que fomenta la adopción pero limita la monetización directa. Se espera una transición hacia licencias comunitarias con acuerdos de reparto de ingresos para mejorar la rentabilidad. La expansión internacional, especialmente en mercados no estadounidenses, es clave. El enfoque empresarial global está cambiando de maximizar tokens a priorizar el ROI. Plataformas como AWS Bedrock y Gemini ya ofrecen modelos chinos. Para identificar a los ganadores a largo plazo, Goldman Sachs aplica un marco basado en capacidad de fijación de precios, ventaja en costes y solidez financiera. En modelos de texto, Zhipu y DeepSeek tienen la posición más fuerte. En multimodal/generación de video, ByteDance lidera con Seed, seguido por Kuaishou y MiniMax. Se mantiene una calificación de compra para MiniMax y Kuaishou.

链捕手Hace 4 hora(s)

Informe Profundo de Goldman Sachs: ¿Quién será el ganador a largo plazo en la industria china de modelos de IA grandes?

链捕手Hace 4 hora(s)

Trading

Spot
活动图片