Japanese Equities Signal & Auto-Trading System - Sell Project

A complete, working codebase for a Japanese equities signal and automated trading system. It fetches daily prices, computes technical signals, applies risk rules, and places real orders through a Japanese broker API. This is sold as a CODEBASE AND INTEGRATION ASSET, not as a profitable trading strategy. Please read the honest disclosure below before contacting me. === TECH STACK === Frontend: React 18, TypeScript 5, MUI v6, Vite 5, React Router v6, Zustand, React Query, Recharts Backend: Python 3.11, FastAPI, SQLAlchemy 2.x, APScheduler, yfinance, pandas-ta Database: PostgreSQL. Infra: Vercel, Render, systemd daemon with watchdog. === MEASURED SCALE (not estimated) === 53 API endpoints / 14 screens / 5,960 lines of Python / 4,840 lines of TypeScript / 150 tests (pytest 137, Vitest 13) / 128 commits === FEATURES === Daily price ingestion (77 symbols, 38,000 rows of time series). Technical signal generation. Automated order placement during market hours. Live broker integration via kabu STATION API. Backtesting engine with result comparison. Strategy lab for parameter sweeps. Risk management: position limits, per-symbol exposure cap, portfolio loss cap, ATR-based stop loss. Japanese market holiday calendar including substitute holidays. Desktop and email notifications with offline queueing. systemd daemon with self-recovery watchdog. === WHY THIS HAS VALUE === The kabu STATION API (au Kabucom Securities) is hard to work with. It can only be reached through a bridge application that runs locally. Tokens expire every trading day, so you have to design a re-login workflow. Fee conditions (SOR routing) and order-type behaviour are not fully documented. Getting from zero to a real executed order takes meaningful time. This codebase is already past that point, with 24 real-money fills. It also contains Japan-specific logic: holiday and substitute-holiday handling, round-lot sizes, and intraday session timing. === HONEST DISCLOSURE - PROFITABILITY === I am not selling this as a money-making strategy. My own real-money results: capital deployed JPY 302,000, returned JPY 295,628, NET JPY -6,372. Realized P/L on sales (FIFO) was JPY +9,085 across 24 fills. My own estimate of expected value is about JPY 5,000 per month. In backtests, the same strategy with the same capital varied by 13x depending on the period (JPY 13,298/month in 2025 vs JPY 1,092/month in 2026). I do not guarantee that this strategy is profitable. Buy it for the implemented foundation - broker API integration, backtesting infrastructure, risk management, daemonized operation. The strategy logic is designed to be replaceable. === KNOWN LIMITATIONS (all disclosed) === 1. If the host PC sleeps, the process freezes and the whole trading day is skipped. The process does not die, so neither systemd nor the watchdog detects it. Unresolved - moving to a server should fix it. 2. kabu STATION tokens expire daily. Manual login before market open is part of the operating procedure. 3. The 11:30 schedule slot structurally cannot place orders; the first practical slot is 12:30. 4. Deployments are currently down. Backend (Render) is suspended, database (Neon) is over its storage quota. Restoring requires a redeploy and a DB plan change. 5. Some old Vercel frontend URLs return 404. === WHAT IS INCLUDED === Full source (backend, frontend, operational scripts), documentation (requirements, deployment guide, operation manual), backtesting infrastructure and verification scripts. === WHAT IS NOT INCLUDED === My brokerage account, the API usage contract, accumulated price data, and my personal trading records. === REQUIREMENTS TO RUN IT FOR REAL === Python 3.11, Node.js 20+, PostgreSQL, and - for live orders - an au Kabucom Securities account with kabu STATION API access approved. === TERMS === Payment via escrow, Wise, or Stripe. No earn-outs or revenue-share arrangements. Communication in English is BY TEXT ONLY. I do not take voice or video calls. I answer written questions in detail, usually within 24 hours. Handover: repository transfer plus a written walkthrough. I will answer questions by text for 14 days after transfer at no charge.