There are so many issues these days that it's hard to figure out which ones are going up and down and why, so I made this.
The data used is
Content extracted from analysis of YouTube stock expert video captions
3 days worth of stock related news data
S&P500, SOX, Nasdaq, KOSPI, KOSDAQ minute, daily, and weekly chart data
Market briefing data released at 6 am and 4 pm every day (databaseized)
AI market situation analysis (there are some that run hourly, and I use those too)
Fluctuations in raw materials, interest rates, and sectors
Reference indicators such as VIX, vkospi, McClellan index, Hindenburg signal, etc.
After RAG processing the content, I create a prompt and send it to qwen3.6 vllm to receive a response.
I fell asleep last night after asking for the task to be created and implemented, but Telegram notified me that it was completed after automatic testing and agent feedback even fixed bugs.
The first result looks good so I think I'll have fun using it for a while.
