Stock Investment AI
A comprehensive educational and paper-trading platform designed to demystify the stock market through real-time AI integration.
An interactive learning environment that pairs live OHLCV market data with on-demand generative-AI explanations. Beginners drop into a tutoring layer; experienced users get a paper-trading sandbox, a rule-based strategy builder, and a structured reader for market literacy — all in one unified dashboard.
The brief.
Over 60% of American adults hold equities, yet a wide gap separates participation from proficiency. Most retail traders fail to beat the 10% annualized return offered by passive benchmarks like the S&P 500 — the barriers to trading have collapsed, but the barriers to financial literacy and strategic execution remain formidably high.
There is also a disconnect between how institutional players actually function and what ordinary people know about the stock market. The cultural flashpoints — meme stocks, The Big Short, public perception of figures like Warren Buffett — sensationalize the activity rather than teach it. The team set out to move users beyond vibe-based investing toward a high-fidelity sandbox where passive observers can transition into informed strategists.
Target: Newer traders — people just starting to learn the market, recent entrants to the workforce, and younger investors looking for a vertically integrated training ground that combines learning, analysis, and simulated execution in one place.
By integrating real-time market data with on-demand generative AI explanations, we allow users to deconstruct complex technical indicators and market movements as they occur.From the write-up
The landscape.
| Tool | Approach | Weakness | Our edge |
|---|---|---|---|
| Investopedia | Editorial articles + paper-trading game | Mixes educational content with ads and news; paper-trading game omits liquidity and order realism | Tutorials are first-class and ad-free; paper-trading runs on Alpaca for realism |
| Schwab / Robinhood explainers | Embedded learn-pages on broker sites | Hidden inside the broker UI, separated from the chart the user is staring at | Explanations sit one click away from the live chart — no context switch |
| Composer / NinjaTrader | Professional strategy builders | Subscription fees, steep learning curve, professional-grade complexity | No sign-up, no fee, fast iteration loop on rule-based strategies |
The core differentiator is embedding generative AI directly into the interface. Clicking any on-screen element produces an immediate, context-aware explanation — turning a static dashboard into an active learning environment, and turning learning the stock market into something genuinely fun.
The system.
The platform is a browser-based dashboard with five user-facing areas: a home screen, an interactive charting workspace, a paper-trading workspace, a rule-based strategy builder, and an educational reader. The goal is to keep learning, market observation, strategy testing, and simulated execution in one interface so users can move from concept to test against real market data without switching tools.
Navigation is tabbed: Home, Charts, Paper, Strategy, Reader — plus light/dark theme control and a lightweight login form. Each tab acts as a focused tool while shared styling variables and API utilities keep the experience consistent across the platform.
The Charts tab is the core market-analysis workspace. It uses lightweight-charts to render a TradingView-style candlestick view with volume bars, crosshair interaction, dynamic history loading, and periodic background refreshes. A central feature is AI-assisted technical analysis: clicking AI Analyze captures only the currently visible chart window, computes summary metrics (start/end price, percent change, period high/low, volatility, average volume), and routes them through an OpenRouter-hosted LLM. The model is grounded in exactly the data the user is looking at — never asked to reason about an ambiguous ticker in isolation.
The implementation.
The system follows a clear front-end / back-end split. The React/Vite front-end owns the user interface, chart rendering, tab state, theme state, and client-side interaction. A FastAPI back-end centralizes the integrations: it pulls historical OHLCV data, talks to Alpaca for paper trading, runs the rule-based backtesting engine, and proxies LLM calls through OpenRouter.
The Strategy tab is a rule-based builder and backtester. Users assemble IF/THEN rules over indicators (SMA 20/50/200, EMA 20, RSI 14, MACD, Price) with conditions, comparison values, and actions (Buy, Sell, Close, Do Nothing). Submitting a backtest sends ticker, date range, starting capital, and rules to the FastAPI backend, which evaluates the rules in a pure-Python engine and returns performance metrics, an equity curve, and a trade log.
The Paper tab connects the analytical work to simulated order execution via Alpaca. The Reader tab provides the structured educational article surface — a table of contents with section-level navigation, with deep-link entry from the chart workspace so a beginner can jump from an unfamiliar concept in the chart back to its explanation in one click.
The evidence.
Limits & next.
Limits
- Volume-profile estimate falls back to a local approximation when Alpaca market data is unavailable.
- Rule-based strategy builder currently supports a fixed indicator catalogue (SMA/EMA/RSI/MACD/Price).
- Reader content is hand-curated; not yet auto-augmented from live filings.
Next
- Expand the indicator library and let users compose custom indicator math.
- Bring the LLM closer to the strategy builder so it can critique a draft rule set before backtest.
- Wire the Reader so chart clicks deep-link into the matching tutorial section automatically.