AI-Powered Stock Analysis Tool Development

Project Overview

I am building an AI-powered stock analysis tool that shows:

Real candlestick chart
AI-generated future chart (5–10 days forecast)
Up/Down probability (%)
Brief explanation behind the prediction
Basic technical indicator analysis

This is an MVP. I already have a Figma prototype.
Your job is to turn it into a working product.

Work Required (MVP)
1. Backend (Python + FastAPI)

Fetch OHLCV data from Yahoo Finance (expandable later)

/ohlcv endpoint for chart data

/forecast endpoint that returns:

5–10 day forecasted prices

up/down probability

simple explanation (“RSI recovery + EMA trend shift”)

synthetic future candles

2. Forecast Engine

Use Python ML models (no deep learning for MVP):

ARIMA or XGBoost-based prediction

Technical indicators from OHLCV:
RSI, SMA, EMA, MACD, Bollinger, ATR, volume features

Output:

future price points

probability of trend

confidence score

JSON with future candles

Target:
~70% directional accuracy in walk-forward evaluation on selected stocks.

3. Frontend (React or Next.js)

Dual-chart layout:

Left: Real candlestick chart

Right / Overlay: AI future chart (different color)

Ticker search

Timeframe buttons

Loading states & clean UI (matching my Figma)

4. Backtesting Module

Walk-forward validation (rolling window)

Output directional accuracy %

Simple backtest with P&L + transaction costs

5. Deliverables

Complete source code (frontend + backend + ML)

GitHub repo

Docker instructions

README with setup

Deployment on Render / Railway (if possible)

Requirements

Must have real experience in:

Python, FastAPI

Pandas

Time-series forecasting (ARIMA / XGBoost)

Technical indicators

Candlestick charting libraries (Plotly or TradingView Lightweight Charts)

Walk-forward evaluation

Clean modular code

Nice to have:

News sentiment

LSTM / advanced models

Deployment experience

How to Apply

Send:

Your previous time-series or stock-related projects

A short explanation of how you will achieve ~70% directional accuracy

A screenshot or code snippet of any chart/forecast you’ve built before

Your timeline + cost estimate

Important: Mandatory Paid Test Task (48–72 hours)

Before full hire, complete a small paid test:

Fetch AAPL or RELIANCE 6 months OHLCV
Plot candlestick chart
Add 10-day ARIMA/XGBoost forecast overlay
Provide 1 walk-forward accuracy number

This is to verify real skill and avoid fake freelancers.

Payment Terms (STRICT)

0% upfront

Payment is milestone-based only

Each milestone is released after successful testing

I will NOT pay for code that does not run locally

Milestones
1. Backend + OHLCV API (20%)
2. Forecast Engine + API (30%)
3. Frontend Dual Chart (30%)
4. Backtest + Deployment + Documentation (20%)
NDA

Freelancer must agree that:

All code belongs to me

No part of the project can be reused or shared

All designs, concepts, and algorithms are confidential

Red Flags (Auto-Reject)

Your proposal will be rejected if you:

Ask for upfront payment
Guarantee 100% accuracy
Can’t show previous ML or stock work
Avoid the paid test task
Refuse to sign NDA Deliver zip files instead of GitHub
Say “I’ll learn as I go”