Skip to content
FSS Knowledge Base

FinGPT: AI Just Got Serious About the Stock Market

Section titled “FinGPT: AI Just Got Serious About the Stock Market”

Until now, asking ChatGPT to summarize earnings reports or distill Seeking Alpha articles felt like a party trick.

But things just changed.

Non-members can read here or subscribe to stay updated whenever I publish.

Over a year ago, I casually wrote: “This is it! #1 — Turks are finally putting AI to proper use.” It was meant as a lighthearted nod to our local creativity with LLMs — from voice-cloning actors to hacked-together mini copilots.

What I didn’t expect? An avalanche of impressions… Since then, I’ve documented more than 30 “This is it!” moments:

🏥 AI therapists in Turkish hospitals, 🚚 GPT-powered supply chain bots, 🎧 AI musicians and cloned voice actors, 🌾 AI farmers, 🧮 AI risk analysts, 🧑💻 and developers everywhere building magic from scratch.

Fast forward to today, the series has racked up 400,000+ impressions across platforms and evolved into a kind of living documentary of how AI is crashing into everyday life — one field at a time. Now here we are at #10. And AI just entered the stock market — for real.

Columbia researchers recently built a production-ready, end-to-end LLM-powered trading system that:

  • Reads real-time news, earnings releases, even Reddit threads
  • Gauges sentiment and named entities
  • Applies technical indicators like RSI, SMA, and Stochastic Oscillator
  • Fires actual trading signals — no human analyst involved

But here’s the kicker — the core engine is open-source: FinGPT, available on both Hugging Face and GitHub:

Find the original FinGPT paper from 2023 here -> https://arxiv.org/abs/2306.06031

🚀 Key features:

  • ✅ Combines LLM-based sentiment with technical indicators like RSI, SMA, and Stochastic Oscillator
  • 🧠 Uses a finance-specific LLM, FinGPT, fine-tuned for real market language
  • 📊 Achieved +700% improvement in Sharpe ratio over traditional technical-only trading
  • 💡 Shows 30–60%+ win ratio uplift for stocks like $TSLA, $AAPL, and $AMZN
  • 🌐 Deployed live via Kubernetes, Colab, and FastAPI

So yes, you can:

Ask an LLM to monitor earnings releases, Reddit, and Bloomberg at once. Generate signals more robust than just RSI threshold. Build a backtested system that actually beats dumb money.

📉 Still on Turkiye’s BIST? Hope not, honestly, don’t. Too many whales, too much noise.

📈 Instead, watch the S&P 500 as the rate cut cycle begins. The stage is set.

This is no longer theoretical.

🔗 Full paper here

📦 GitHub: https://github.com/Ronitt272/LLM-Enhanced-Trading

#ThisIsIt #LLMTrading #FinGPT #QuantFinance #SP500 #NotFinancialAdvice

Best

Altan

P.s.

For the previous post in this series check this link ->## 🧠 This is it! — #9 The First AIs With 130+ IQs Started Popping Up. Seriously!

Not long ago, language models were clever assistants — parroting back facts, drafting emails, maybe summarizing PDFs…

Section titled “Not long ago, language models were clever assistants — parroting back facts, drafting emails, maybe summarizing PDFs…”

[

medium.com

](https://medium.com/@atabarezz/this-is-it-9-the-first-ais-with-130-iqs-started-popping-up-seriously-06d7328aaff4?source=post_page-----6e99341a6e1d---------------------------------------)

Also P.s.

Among the 100+ use cases identified by HBR, this sits at number 6 (investment ideas), 32, 49, 73 (buy and sell) and 94.

Ideas for the future of society, #AI and #innovation — connecting big-picture strategy with real-world execution. x-SAS, x-Experian, x-GarantiBBVA, METU, YTU

Responses (1)

Section titled “Responses (1)”

Talbot Stevens

What are your thoughts?

Totally agree 100%
Currently working on AI trading system using trend following, mean reversion or statistical arbitrage strategies optimized for use over multiple time frames from intraday to quarterly.
Great time to be a software developer that knows how to trade too!!!

3

More from Altan “Atabarezz”

Section titled “More from Altan “Atabarezz””
Section titled “Recommended from Medium”

[

See more recommendations

](https://medium.com/?source=post_page---read_next_recirc—6e99341a6e1d---------------------------------------)