I’ve been feeling consistently down lately. Even the project I was most passionate about at work is becoming a source of doubt, and my relationships with people are just adding to my fatigue. It’s not that the company has treated me particularly poorly, but things aren’t going smoothly, and I feel like I’m the one absorbing all the negative vibes. It’s exhausting.

I’ve started going to the hospital because my body isn’t doing well. My chest feels tight, and I have headaches, but all the tests say I’m “normal.” They suspect it’s psychological, but when you can actually feel the pain, it’s confusing whether to be relieved or not. While my primary doctor was away, I met a Korean doctor at the same hospital, which allowed me to explain my condition in more detail.

Here’s what my life looks like these days:

• Going to the hospital
• Preparing for a job change
• Meeting people and going out
• Writing with ChatGPT
• Creating a cryptocurrency automated trading program

Cryptocurrency Automated Trading: Restarting the Experiment

This isn’t my first attempt. I first tried it during the pandemic out of sheer boredom. I used historical data to analyze various strategies and validated their performance through backtesting. The system generated buy/sell signals based on the most profitable strategy.

The result was a failure. It turned out that just holding onto the assets would have been more beneficial, and my program was structured to automatically incur losses. But at least I learned a lot by building everything myself. Considering time efficiency, investing in stable ETFs was a better option.

Moreover, the Robinhood API I used back then wasn’t officially sanctioned, which made me anxious about my account getting suspended. As work got busier, the project naturally faded away.

Fast forward a few years, and things have changed. Now there’s ChatGPT. Just as smartphones changed the world, LLMs are doing the same now. It may not have emotions, but it produces results as if it thinks, perhaps even more logically than humans.

A Fresh Start: New Tools and Structure

ChatGPT introduced me to a platform called Alpaca, which offers a legitimate API and even allows for paper trading. It suggests new strategies and assists with coding. Recently, I purchased a cloud server on AWS Lightsail. Since it’s a 24/7 server, I can run the system there for true automated trading.

So, I’m keeping Robinhood for stock trading and running a cryptocurrency—specifically dollar-based Bitcoin—trading simulator. If the results look promising, I’ll switch to real trading. Since a lot of my work involves data-driven predictions, this project serves as a kind of refresh for me.

System Overview

1. Start with $1,000,000 in virtual assets
2. Collect recent 30-day Bitcoin 5-minute interval data (around 8,640 data points)
3. Backtest using 25 common strategies + 5 machine learning strategies
   1. Common strategies: SMA, RSI, MACD, Bollinger Band, etc.
   2. ML strategies: Probability-based, LSTM, Random Forest, XGBoost, Ensemble
4. Calculate each strategy’s return → Generate buy/hold/sell signals
5. Heavily weigh against low-performing strategies
6. Weighted average of all opinions → Determine final strategy
7. External factors: Incorporate Bitcoin Fear & Greed Index
8. Execute strong buy/sell if strategy aligns with index, otherwise take a cautious approach

Next Steps

• Integrate with Telegram bot: Real-time alerts for trades
• Incorporate news analysis: Additional strategy determination through article analysis beyond fear index

Current Status

…Currently, all I’m getting is hold signals, so there’s no trading going on.
Honestly, I have no idea why this is happening, lol.