Đức nguyễn

Gần 20 năm kinh nghiệm đầu tư và giao dịch tại thị trường chứng khoán Việt Nam

[そう｜Claude Codeで始めるAI自動化 @so_ainsight]

Claude x トレーダー、えぐい。

AnthropicのAI「Claude」が、株式・暗号資産のチャートツール「TradingView」をマウス操作ゼロで動かしている事例が海外で話題に。

・RSIや出来高などの条件を自然言語で指示→14秒で6銘柄を自動発見・チャート表示
・過去の売買をリプレイし、損益をローソク足単位でリアルタイム計算
・オリジナルの分析インジケーターを40秒でプログラミング・表示

など、「チャートを見てから考える」作業全体をClaudeに丸投げする仕組みです👇

[Deedy @deedydas]

Claude Fable 5 is by far the most ridiculous model that makes me genuinely afraid for the future of software engineering.

I compiled the top 10 most unbelievable things I've seen Claude Fable 5 do today:

— Migrate a 50M line codebase from Stripe in a day (humans take 2mos)
— Draw amazing 3D graphics a) Boeing 747 b) space simulations with >5000 objects c) Minecraft roller coasters d) full photorealistic forest scenes e) NYC skyline f) stormy clouds)
— One-shot Pokemon FireRed the game
— Optimize a real world proprietary interaction net evaluator 10x more than the next best model, gpt5.5

AND it's about the same price as GPT 5.5 ($10/M input, $45/M output) vs Fable 5 ($10/M input, $50/M output) and 6x cheaper than GPT 5.5 Pro.

[karla @karlarboledas]

¡CLAUDE CODE AHORA OBTIENE DATOS EN VIVO DE MÁS DE 17.000 ACCIONES, CRIPTOMONEDAS Y ESTADOS FINANCIEROS EN SEGUNDOS! 📊

Un comando. 60 segundos. Listo.

Aquí tienes la configuración exacta paso a paso:

Paso 1: Configura la conexión
Abre Claude Code y pega el siguiente comando:
claude mcp add --transport http financial-datasets https://t.co/DA6F99tee8

Paso 2: Autentícate
Escribe /mcp dentro de Claude Code y completa el rápido proceso de OAuth en tu navegador.

Tip: Puedes verificar que la conexión está activa en cualquier momento usando: claude mcp list

Paso 3: Empieza a consultar
Ya puedes hacerle preguntas financieras avanzadas en lenguaje natural, como por ejemplo:

"¿Cuál es el ratio P/E actual y la capitalización de mercado de Apple?"

"Muéstrame el estado de resultados de Tesla de los últimos 4 trimestres."

"¿Cómo ha variado el precio de Bitcoin durante el último año?"

Y eso es todo.

De la noche a la mañana, Claude Code obtiene acceso directo a datos de mercado reales: más de 17.000 acciones, informes de ganancias, balances generales, flujos de caja y precios de criptomonedas.

A los analistas que pagan $24.000 al año por una Terminal Bloomberg no les va a hacer ninguna gracia que esto exista. Antes, para lograr esto necesitabas software carísimo, lidiar con APIs complejas o perder horas saltando entre múltiples fuentes.

Hoy, solo necesitas un comando y un minuto de tu tiempo.

Los analistas cuantitativos, investigadores y gestores de portafolios que aprendan a combinar la capacidad de razonamiento de Claude Code con este acceso a datos en vivo tendrán una ventaja competitiva brutal en el mercado.

🔖 Guarda esto en favoritos para tenerlo a mano antes de hacer tu próximo análisis financiero.

[gus @igus_ai]

Cancelé $2.000/mes en suscripciones de Trading

Reemplacé casi todo por repositorios Open-Source 100% gratis

Este es el stack completo:

1. TradingView Pro ($30/mes) → lightweight-charts

14K estrellas. Creado por el propio equipo de TradingView. 45KB. Gratis.
> https://t.co/VqpSa8RNuR

2. Bloomberg Terminal ($2.000/mes) → fredapi + Claude

Acceso a todos los datasets macroeconómicos publicados por la Fed mediante API gratuita
> https://t.co/1dvvJRkXVB

3. Plataforma de backtesting ($100/mes) → prediction-market-backtesting

Fork de NautilusTrader con adaptadores para Polymarket y Kalshi
> https://t.co/wzFhoGQNbG

4. Ingeniería inversa de estrategias → polybot

Infraestructura de ejecución y datos de mercado con paper trading.
Kafka, ClickHouse y Grafana como pipeline completo de analíticas
> https://t.co/x3rufeBuyX

5. Paper trading para agentes IA → polymarket-paper-trader

Order books reales, modelo exacto de fees y tracking de slippage tu agente de Claude recibe $10K ficticios para operar
> https://t.co/kp9IZyacpF

6. Ahorro de tokens → rtk

Proxy CLI que reduce entre un 60-90% el consumo de tokens en Claude Code
escrito en Rust, binario único y compatible con 10 herramientas IA
> https://t.co/9n4E6OdxA6

7. Claude Code ($200/mes) → goose

35K estrellas. Desarrollado por Block (Jack Dorsey). Escrito en Rust. Funciona con cualquier LLM y ofrece un loop completo de agentes IA
> https://t.co/S8SDZjNbwz

Antes: +$2.600/mes
Ahora: prácticamente $0

Guárdate este post, me lo agradecerás. 🔖

[Marlow @marlowxbt]

A 14 year old in China learned Claude Code from YouTube tutorials over summer break. Watched the same 3 videos over and over until he understood how prompts work. By September he was building things his CS teacher couldn't.

First week back at school the teacher gave the class an assignment: build any simple program using Python. Most kids made calculators. One kid made a to do list. The usual.

This kid built a full multiplayer math game on a touchscreen. Tug of war. Two teams stand on each side of the screen. A math problem appears for each team. Solve it faster and your team pulls the rope. Wrong answer and the other side gains ground. Timer counting down. Animated characters pulling. Kids screaming.

The teacher watched the demo and went quiet. Then asked: how long did this take you?

Him: one evening. I described what I wanted to Claude Code and it wrote most of it. I just fixed the parts that didn't look right.

The teacher had spent three weeks building a quiz app for the same class and it still had bugs. This kid built something better in one evening by talking to an AI.

His mom filmed the class playing it. 17 seconds. Four girls jumping at the screen trying to solve 5+10 and 6+8 before the timer runs out. Team 1 leading 3 to 1. The rope pulling left. Kids going crazy.

She posted it to a parents' group. Someone shared it publicly. 2 million views in a week.

Schools across the province started asking for the code. The kid put it on GitHub. 400 forks in a month. Teachers in three countries now use it in their classrooms.

The CS teacher uses it too. In his own class. The game his 14 year old student built in one evening that he couldn't build in three weeks.

He never told the other students who made it. Just said he found a good educational tool online. The kid sits in the second row every class watching his teacher demo his own game and pretends he's never seen it before.

His classmates learned Python by reading textbooks. He learned Claude Code by watching YouTube at 2am. They write 50 lines in a week. He ships full apps in an evening.

The teacher gave him an A on the assignment. Highest grade in the class. Then asked him after class: can you teach me how you use that AI tool?

A 14 year old is now teaching his CS teacher how to use Claude Code. The teacher has a degree in computer science. The kid has a YouTube history full of Claude Code tutorials and a GitHub with more forks than his teacher's entire career.

Same classroom. Same subject. Same Python. One learned it from a university. The other learned it from a 12 minute YouTube video and an AI that doesn't care how old you are.

The math game is still running in his school. His teacher still presents it as a tool he found online. The kid still doesn't correct him. He said it's fine, the game works better when nobody knows a 14 year old made it. Teachers take it more seriously that way.

[FattyNatsu @FattyNatsu]

$150k–$300k
That's all it takes to be richer than 99% of Vietnam 🇻🇳

102 million people. One of Asia's fastest growing economies.

Here's who actually has the money:
▲ 8 billionaires - $1B+
▲ ~750 ultra-wealthy - $30M+
▲ ~5,500 people - $10M+ (0.005% of the country)
▲ ~100k millionaires
▲ Top 1% starts at $150k–$300k
▼ Median adult: ~$7,000

The wealth is growing fast. The pyramid isn't getting flatter.
Most people are closer to $7k than $150k.

*Estimates via Knight Frank 2024, Forbes 2026, UBS, Henley & Partners

[wincy.eth @gusik4ever]

the fastest growing GitHub repos in finance this week:

1. shiyu-coder/Kronos (+6.5K ★)

first open-source foundation model for financial candlesticks. trained on 45+ global exchanges. predicts OHLCV candles as tokens — literally GPT for price charts. accepted at AAAI 2026.

2. virattt/ai-hedge-fund (+4.9K ★)

a team of AI agents simulating Buffett, Munger, Ackman, Cathie Wood and others. each agent runs its own strategy, a Portfolio Manager makes the final call. one of the most viral finance repos right now.

3. TauricResearch/TradingAgents (+~3K ★)

multi-agent LLM trading framework. fundamental analyst, sentiment analyst, technicals, risk manager — all working together. supports GPT-5.x, Gemini 3.x, Claude 4.x, Grok. built by UCLA/MIT researchers.

4. ZhuLinsen/daily_stock_analysis (+~2K ★)

LLM stock analyzer for US, A-share and H-share markets. auto-builds a daily decision dashboard with exact entry/exit levels. pushes to WeChat/Telegram/Discord/Email via GitHub Actions. zero cost, zero server.

5. hsliuping/TradingAgents-CN (+~1.5K ★)

Chinese fork of TradingAgents. fully localized for A-share markets (Shanghai/Shenzhen), Chinese data sources, and domestic LLMs. 5.1K forks — very active community.

6. OpenBB-finance/OpenBB (+~1K ★)

open-source Bloomberg alternative. stocks, crypto, options, derivatives, fixed income — one platform. integrates with AI agents via MCP. 66K total stars and still climbing.

7. freqtrade/freqtrade (+~700 ★)

free, open-source crypto trading bot in Python. supports all major exchanges, full backtesting, strategy optimization, Telegram control. release 2026.3 just dropped.

8. AI4Finance-Foundation/FinGPT (+~500 ★)

open-source financial LLMs trained on real market data — news, filings, earnings. built for sentiment analysis and robo-advisors. models on HuggingFace, ready to deploy.

9. juspay/hyperswitch (+~400 ★)

open-source payments router in Rust. one API to connect Stripe, Adyen, PayPal and 50+ providers. smart routing, high performance, built for fintech scale.

10. microsoft/qlib (+~350 ★)

Microsoft's AI quant investment platform. covers the full pipeline: alpha seeking, backtesting, model training, live trading. supports ML/DL, RL, and auto-quant.

bookmark this and start today.

[Hanako @hanakoxbt]

an ex-Citadel quant told me a salary is a short position on your own time

we were at a dinner in new york. friend of a friend. he asked what i do.

"software engineer. $4,200 a month"

he said one simple thing

"you're mass-selling the most valuable asset you have. your hours. at a fixed price. with no upside. that's a short position on your own life"

i didn't have a response.

then he showed me his phone. a terminal. live trades. +$39,453 running for 6 weeks.

"i built this with Claude Opus 4.7 and one open source dataset. took a weekend. haven't touched it since"

i asked how.

"gave Opus one repo and one prompt. find who wins on prediction markets. find why. copy their edge"

https://t.co/xywgRNlvyc

2,900 stars. 36GB. every trade on Polymarket and Kalshi. every wallet. every resolution.

Opus read the whole dataset friday night.

by sunday it had a live terminal scanning 1,400 markets per hour. 8 detectors running in parallel. entering when void score exceeds 90. exiting when it drops below 75. no human override.

> Senate filibuster. locked at 42c. now 98c. +$1,133.
> Neuralink approval. entered 52c. now 98c. +$1,396.
> Fed 0% rate. entered 35c. now 97c. +$1,147.
> BTC 250K. entered 8c. now 98c. +$1,258.
> Megaquake. entered 6c. now 97c. +$1,039.

the part that broke something in me: latency arbitrage. Polymarket updates in 19ms. Kalshi in 63ms. the bot trades the 44ms gap between them. automatically. every time.

"we had four people at Citadel doing what your bot does with latency alone"

copy mirror tracking 6 positions:

> Senate fili +$1,252. void signal. 10%.
> BTC 250K +$958. zero signal. 2%.
> Neuralink +$1,384. sign signal. 9%.
> Fed 0% +$863. fill signal. 8%.
> Megaquake +$1,070. bridge signal. 18%.
> META AI +$979. gas signal. 5%.

972 trades. 83% win rate. kelly f+ 0.101. avg profit $10.65. drawdown -1.4%.

maker rebate tracker collecting $203 across four platforms while the bot trades. Polymarket +$114. Kalshi +$40. dYdX +$28. Betfair +$21.

$1,800 seed. +$39,453.

i quit the next morning. one email. "i'm done. thank you for everything."

copytrade setup: https://t.co/PTZuvewZE6

texted him a week later. "you ruined a perfectly stable career"

he replied "no. i closed your short position"

[Noisy @noisyb0y1]

A regular American developer bought $1,400 worth and stacked seven Mac Minis on top of each other and connected them with metal cables.

Neighbors thought he was building a mining server. His wife thought he'd lost his mind. He just didn't want to pay $15,000 a month for a dev team.

On the screen - a diagram. Seven Mac Minis connected via Ethernet working as one machine. EXO framework distributes tasks between them automatically. 11.44 TFLOPS each. Together - more than most cloud servers that companies pay thousands for every month.

He paid $1,400 for the hardware once.

38 agents from GitHub, 156 skills. A system that learns from session to session and in two weeks writes code just like he does - but seven times faster because it runs on seven machines in parallel.

A task that took a junior dev 10-12 hours - the tower closes in 20 minutes.

One founder with this setup ships a product like a team of eight people.

For $20 a month instead of $120,000 a year.

This 7 Mac Mini setup helped him win the Anthropic hackathon and make $26,000 without a team.

[kepo @kepochnik]

9 GitHub repo for trading on Prediction Markets:

1. Analysis: https://t.co/cprtuRZn1p
data analysis for Polymarket and Kalshi

2. Polymarket Agents: https://t.co/cVhDaXbZHe
framework for building AI agents with API integrations and RAG support

3. CLI: https://t.co/yD6NInwWKU
browsing markets placing orders managing your wallet

4. pmxt: https://t.co/ZbIwzgJ7on
unified library for working across multiple prediction market platforms

5. CLOB Client: https://t.co/z5zjFCVmc6
place orders and manage positions through a simple API

6. TRUMP CODE: https://t.co/AAWHEe25iy
analyzes Donald Trump posts and looks for related arbitrage opportunities

7. Polyseer: https://t.co/2IXFK5P3qC
AI tool for deep prediction market analysis using Bayesian evidence aggregation across multiple data sources

8. PolyClaw: https://t.co/jbiiRDWl0Q
CLI plugin for trading on Polymarket with AI-powered hedge discovery

9. Polymarket Data: https://t.co/EiSoSKF6Nv
107 GB dataset containing 1.1B Polymarket trades from the blockchain for research

completely free

[Noisy @noisyb0y1]

Chinese student spent $4,000 on two Mac Studios and a Mac Mini and built a Claude + Obsidian bot

Claude and Obsidian scan hundreds of factors simultaneously. a human physically can't work with more than 2-3 processes at once. the machine does 12 microprocesses in seconds.

three commands:
/wiki - initialization
/save - save the conversation
/autoresearch - deep research on a topic with automatic citations.

stuck stickers on them: UI/UX. DEV. ADMIN

Three computers do one thing: Claude + Obsidian pulls live data that the developer feeds in and builds the product. updates every 1-3 hours running 24/7

[Quant Science @quantscience_]

Machine learning techniques and data for stock market forecasting.

(A free 41-page PDF) 👇

[Elja @Eljaboom]

RIP TRADERS!

CLAUDE IS NOW RUNNING TRADINGVIEW LIVE FROM THE TERMINAL.

PINE SCRIPT. FUTURES. LEVELS. EVERYTHING. AUTOMATICALLY.

TRADERS ARE OFFICIALLY COOKED.

[Axel Bitblaze 🪓 @Axel_bitblaze69]

Dear Traders,

Claude code can now connect directly to your TradingView charts

It reads your charts, indicators, writes Pine Script for your trading strategies, compiles it and fixes errors.. all automatically

Not a browser extension.. A direct connection to your live charting platform

▫️Here's how it works and how to set this up:

> How it works:

TradingView Desktop runs on Electron (Chromium). Chromium has a built in debug interface called Chrome DevTools Protocol. It's off by default. One flag turns it on.

That's the door. Claude Code connects through it via an MCP server running locally on your machine.

Nothing touches TradingView's servers. No data leaves your machine. Everything runs on localhost.

Here's the setup, give Claude code this prompt and it does everything:

"Install the TradingView MCP server. Clone and explore https://t.co/4NApHd4Lyx, run npm install, add to my MCP config at ~/.claude/.mcp.json, and launch TradingView with the debug port."

Once it's done, verify:

"Use tv_health_check to confirm TradingView is connected."

Then just try:

"What's on my chart right now?"

That's the whole setup. Claude installs it. Claude runs it. You just describe what you want.

▫️What Claude can actually see

What Claude can read from your charts:

> Current symbol and timeframe
> Real-time OHLC and volume
> Up to 500 price bars
> All indicator names, IDs, and values
> Pine Script drawings — lines, labels, tables, boxes
> Data window values
> Order book depth
> Strategy tester results and trade list
> Screenshots of any chart region

The most useful part is it reads protected indicators too.

Pass a filter with the indicator name and you get back every price level, every text annotation, full table contents, every drawn zone.

"What levels is my NY Sessions indicator showing?"

it reads them.

"What does the session stats table say?"

it reads that too.All of this data is already on your screen. This tool just makes it readable to an AI.

▫️ The Pine Script workflow

This is the part that changes everything for traders who code:

You describe what you want → Claude writes the Pine Script → injects it into TradingView → compiles it → reads the errors → fixes them → recompiles That loop runs until the script is clean.

The whole time Claude has full context of your chart and every indicator on it.

No more copy-pasting between VS Code and the Pine editor hunting type mismatches.

You can also:

→ Validate indicators (write it, compile it, read values back, check the math)
→ Loop through input parameters and test each one
→ Run the same indicator across multiple symbols and timeframes
→ Read strategy tester metrics after each compile

The actual workflow of verifying your indicators works instead of just assuming they do.

sharing the usecases of this workflow in my next post.. like & RT

[Tips Excel @gudanglifehack]

My father is 83 years old and has studied US stocks for almost 52 years.
He summarized some simple rules for beginners:
1. Price falls 5% → Hold
2. Price falls 15% → Buy 10%
3. Price falls 25% → Buy 25%
4. Price rises 5% → Continue holding
5. Price rises 15% → Continue holding
6. Price rises 25% → Sell 10%
7. Price rises 35% → Sell 20%
8. Price rises 45% → Sell 30%
9. Price rises 60% → Sell 40%
10. Price rises 100% → Sell everything
Discipline + patience = stable long-term growth.

(Save this later).

[Mark Minervini @markminervini]

The market's character is still one of a bear market or cyclical correction; strong open, fade into close and major average living below the 200-day line. Before a reliable bottom can be established, we need to see better price and volume action, including better action from breakout names forming bases.

We are clearly NOT out of the woods yet. The market backdrop is one where sentiment has improved with rising pessimism, but not a full capitulation. The VIX has reached bear warning levels, but remains below true washout extremes. A volatility washout is not required for a bottom, but would add conviction.

Bullish Scenario
--The war ends
--Oil prices recede
--Stagflation concerns ease
--Central banks continuing their easing trajectory

Under this scenario, we would expect:
-A broadening market advance
-Emergence of new leadership from sound bases
-A Follow-Through Day (FTD) on the NYSE and/or NASDAQ confirming institutional buying with little in the way of immediate distribution
-Significant drop in volatility

Bearish Scenario
--The war persists or escalates
--The Strait of Hormuz remains disrupted
--Oil prices make new highs
--Stagflation becomes evident in hard economic data

This would likely result in:
-Limited general market rally attempts with most breakout stocks failing
-Lack of follow-through from breakout names
-Further deterioration in breadth and leadership
-Dearth of setups in buyable position
-Continued elevated volatility and distribution
In that case, sentiment would likely need to reach higher levels of pessimism before a durable market bottom could form. In its absence, and end to the factors that are pressuring the market could cause the market to bottom in less dramatic fashion.

[Reuters @Reuters]

A South Korean plastic film factory has cut production to 20–30% of capacity due to raw material shortages caused by the Iran war and disruptions in global energy markets https://t.co/fyzUViZvtO

[Đức nguyễn @luongnk]

@LeninUGReal lấy dữ liệu từ các nguồn nhất là facebook hoặc trong zalo thì làm thế nào bác

[Paone @paonx_eth]

https://t.co/u7qLqAHTIa