Pillow Investment Method Complete: Fully Automated AI Stock Trading That Makes You Excited for the Morning — Claude Vibe Coding
"Why does my automated trading system collapse in a declining market?" I, too, have wrestled deeply with the same question. Most automated trading logic on the market may appear to generate profits during bull markets, but the moment market conditions change, it loses its ability to adapt. To solve this problem, I combined the expertise I developed through two years of hands-on research with the latest AI technology. In this course, you will learn how to design a 'Transformer-based deep learning model' that simultaneously reads macroeconomic trends and the movements of individual stocks, and how to automatically train and deploy it in a free GPU environment. In particular, by introducing sophisticated, hedge-fund-level data normalization techniques, you will build an 'intelligent comprehensive scoring system' that goes beyond simple indicators by incorporating short-selling balance and order-flow data. On top of that, an ‘AI safety gate’ infused with the insight of a Wall Street trader with 20 years of experience provides the final confirmation from the perspective of a human trader just before a purchase. Going beyond simply buying and selling, you will implement the entire process of an institution-grade algorithmic system with dynamic take-profit and stop-loss logic, eight self-recovery mechanisms, and real-time Slack reporting—all with your own hands.
308 learners
Level Basic
Course period Unlimited
[Update] Replaced news sentiment with the “short-selling balance supply-demand factor” (Lessons 18 and 19 added) + 30% off for 2 weeks
Hello, this is Chialnam 😊
I’ve added two update lectures to the complete version of the Pillow Investment Method. Existing students can watch them right away at no additional cost.
📌 What has changed?
We replaced the 'news sentiment score' that accounted for 20% of the overall score with the 'FINRA short interest supply-demand score. ML predictions, technical indicators, trading volume, ADX, VIX, and the stop-loss logic remain unchanged.
- Lesson 18: Introducing the short interest supply-demand factor (background for replacing news sentiment and designing the days-to-cover score)
- Lesson 19: Introducing a short interest supply-and-demand factor (FINRA data collection development and incorporating a composite score and Slack evidence card)
📌 Why did we change it?
1) The Alpha Vantage news sentiment API used in Lesson 17 became paid-only on October 7, 2026. Calls are rejected from the very first request when using a free key.
2) Instead of paying, we directly measured the effect of news sentiment using two years of data (25 stocks). Stocks with more positive news lagged by 0.7 percentage points after 10 trading days. This is because good news is already priced into mega-cap stocks.
3) Just as the Korean stock section looks at foreign and institutional flows, we compared 22 free datasets in the U.S. that serve the same purpose using the same method. Only short interest passed validation. The top five stocks with the lowest short interest outperformed the bottom five by +3.2%p after 10 trading days, and all four half-year periods showed the same direction.
📌 How was it applied?
- The pipeline automatically retrieves the short interest files published twice a month by FINRA at 9 p.m. on weekdays. No registration or API key is required.
- For each stock, we calculate the "number of days of trading volume needed to buy back the shorted shares (days-to-cover)" and turn both whether it is lower than other stocks and whether it has decreased compared to usual into a score.
- An AI review immediately before purchase and the Slack recommendation rationale card will now include a “supply and demand (short interest)” item.
- You can turn it on and off with a single switch, so if a problem occurs, you can revert to the previous method within one minute. Existing data will not be changed.
📌 What has improved?
- You can continue operating the system without a paid API. There is no additional cost.
- We transformed signals whose effectiveness had not been confirmed into signals that passed validation.
- Since it does not overlap with the price chart, it complements existing technical indicators.
- You can see at a glance in the Slack card why this stock was recommended, including supply and demand.
📌 How existing students can apply
After getting the latest code with git pull, follow the eight steps in the notes for Lesson 19. If you get stuck, please leave a question on the Q&A board.
🎉 Update Celebration Discount (2026.10.20 ~ 11.03, for 2 weeks)
- Complete Pillow Investment Method 30% (132,000 won → 92,400 won)
- Pillow Investment Method (Beginner's Edition) 30% (99,000 won → 69,300 won)
- The Secret of Algorithmic Trading 30% (47,300 won → 33,110 won)
- 50% off the 4-course roadmap bundle (including the Korean stocks course; get the coupon on the roadmap page 👉 https://inf.run/m2EXD)
The discount ends at 11:59 PM on November 3.
※ The verification is a backtest based on 25 mega-cap stocks held for 1–2 weeks over a 2-year period and does not guarantee actual returns.
Thank you.
Best regards, Chialnam




