hengruiyun/AI-Stock-Master

AI Stock Master – What It Is

AI Stock Master is a desktop‑focused platform (Windows / macOS) that combines traditional quantitative stock‑analysis techniques with large‑language‑model (LLM) interpretation. It targets Chinese, Hong‑Kong and U.S. equities and aims to turn raw market data into natural‑language reports and trading signals.

---

Core ideas

Component	What it does	How it’s built
RTSI (Individual Stock Trend Strength Index)	Scores each stock’s trend strength using price/volume trends, statistical consistency, confidence (R²) and a volume factor.	A weighted linear formula whose coefficients (α₁‑α₄) are tuned by machine‑learning on historical data.
TMA (Technical Momentum Analysis)	Gives a single momentum score for an industry sector by fusing RSI, MACD, volume‑profile and other technical indicators.	Similar weighted formula (β₁‑β₅) with adaptive weights learned from back‑testing; includes outlier handling and cyclical adjustments.
MSCI (Market Sentiment Composite Index)	Produces a market‑wide sentiment number from VIX‑style fear index, capital flow, volatility, long/short ratios and NLP‑derived news sentiment.	Linear combination (γ₁‑γ₅) where the news‑sentiment term comes from an LLM‑based sentiment analyser.
LLM interpretation layer	Turns the numeric outputs of the three algorithms into readable analysis, recommendations, and a risk‑score.	Uses a locally‑run LLM via Mini Ollama (a lightweight runtime that can load open‑source models without internet).

---

How a typical analysis runs

1. Data ingestion – market price, volume and technical data are fetched, validated and cached.
2. Algorithmic scoring – RTSI, TMA and MSCI are computed in parallel.
3. Multi‑dimensional aggregation – the three scores are combined into a structured result set.
4. Local LLM – Mini Ollama receives the structured data and generates a natural‑language report, charts and trading signals.
5. Output – the UI shows charts, a risk assessment, and a written commentary that the user can query further.

---

Getting started (as described in the README)

# On Windows/macOS simply double‑click the provided batch file
AI-Stock-Master.bat

The batch script launches the graphical interface; no additional configuration is required because the program auto‑detects and starts Mini Ollama in the background.

---

What you actually get

• Individual‑stock view – trend score, buy/sell signal, risk level.
• Industry view – ranking of sectors, rotation recommendations, suggested allocation weights.
• Market‑wide view – MSCI sentiment index and overall market status.
• Conversational Q&A – ask the LLM about the numbers, ask for “why” explanations, or request a summary report.
• Back‑testing module – run historical simulations and see returns, Sharpe ratio, max drawdown.

---

Limitations & warnings (explicit in the repo)

• The software is research/learning only – it is not licensed as financial advice.
• Historical back‑tests (e.g., 5‑year excess return ≈ 12.3 %) do not guarantee future performance.
• LLM outputs can be wrong or over‑confident; users must combine them with personal judgment.
• Data quality and market extremes can break the algorithms (over‑fitting risk, extreme volatility).

---

Who might find this useful?

• Quant‑oriented hobbyists who want a ready‑made desktop tool that mixes classic technical analysis with AI‑generated commentary.
• Students or researchers exploring how LLMs can be integrated into financial‑signal pipelines.
• Developers looking for a reference implementation that shows Mini Ollama integration and a multi‑algorithm architecture.

---

Quick take‑away

AI Stock Master is a genuine software project that implements three proprietary quantitative indicators (RTSI, TMA, MSCI) and layers a locally‑run open‑source LLM (via Mini Ollama) to produce human‑readable stock‑market analysis. It is positioned as a learning/research tool rather than a production‑grade trading system, and the README makes that distinction clear.