anti-gambling-trader-tw
免費開源的投資反詐、交易統計與自動化交易程式工具:預設 PaperBroker,提供 14 種券商/交易所選項;交易紀錄分析在本機執行。
尚未偵測到更新。GitHub 發布版本後會顯示在這裡。
個人 GitHub 帳號下公開儲存庫的擁有者可以登入並驗證所有權,自行管理介紹,無需管理員核准。
登入並產生驗證檔案,在 30 分鐘內將其提交至此儲存庫的預設分支。其他分支或分叉儲存庫中的檔案無效。然後在下方檢查檔案,即可無需管理員核准直接管理專案。請保留檔案以維持管理權限。
如需登入,將前往登入頁面。登入後請回到此處,再次點擊所選的驗證按鈕。
在專案資料夾中執行 /nomorevibe verify,即可驗證所有權並自行管理介紹。
一分鐘完成驗證Your AI-powered personal wealth advisor. Automated portfolio monitoring, market analysis, and investment briefings delivered via Telegram. Runs on Claude Code with your existing Max/Pro subscription — no API costs.
翻譯準備中
正在載入完整 README。 以下為已儲存的摘錄。
Your AI-powered personal wealth advisor. Automated portfolio monitoring, market analysis, and investment briefings — delivered via a self-hosted web dashboard and Telegram bot.
Runs on Claude Code with your existing Max/Pro subscription — zero API costs for AI.檢視完整 README ↗ ██╗ ██╗███████╗██╗ ██████╗ ██████╗ █████╗
██║ ██║██╔════╝██║ ██╔═══██╗██╔══██╗██╔══██╗
██║ ██║█████╗ ██║ ██║ ██║██████╔╝███████║
╚██╗ ██╔╝██╔══╝ ██║ ██║ ██║██╔══██╗██╔══██║
╚████╔╝ ███████╗███████╗╚██████╔╝██║ ██║██║ ██║
╚═══╝ ╚══════╝╚══════╝ ╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝
Your AI-powered personal wealth advisor. Automated portfolio monitoring, market analysis, and investment briefings — on a self-hosted web dashboard, with an optional iOS app and optional Telegram bot.
Runs on Claude Code with your existing Max/Pro subscription — zero API costs for AI.
Twice a week, Velora collects market data for your entire portfolio, pulls macroeconomic indicators, searches for relevant news, and feeds everything into Claude Opus with a specialized financial analyst system prompt. The result is a comprehensive briefing on a professional dark-themed web dashboard — and, if you want, as a notification on your phone or a Telegram message. Like having a CFA on retainer.
| Category | Highlights |
|---|---|
| Web Dashboard | Dashboard, Portfolio (incl. performance & tax), Market, Briefings, Recommendations, Strategy, Look-through, Chat, Settings. Dark theme, responsive, installable as an app (PWA), multi-language (de/en) |
| AI Chat | Talk to Velora about your portfolio in the browser or the iOS app; it can log trades and orders after your confirmation |
| iOS App (optional) | Native SwiftUI app you build yourself — free Apple account works, push notifications with a paid developer account |
| Telegram Bot (optional) | Briefings, trade logging (natural language), ticker analysis, free-form chat, watchlist |
| AI Analysis | Bi-weekly briefings, monthly reports, on-demand deep-dives, persistent memory across sessions |
| Portfolio Tracking | Multi-account, multi-currency (EUR/USD), tax-loss harvesting, benchmark comparison |
| Recommendations | BUY/SELL/WATCH tracking, automatic outcome detection, hit-rate stats, trade auto-close |
| Data Sources | yfinance, FRED, ECB, Brave Search, Bloomberg RSS, Finnhub, CNN Fear & Greed |
| Deployment | Runs on Raspberry Pi (~90MB RAM), systemd auto-start, cron scheduling |
MARKTLAGE
Relief-Rally (NASDAQ +4.4%, VIX -22%) ändert nichts am Extreme-Fear-Regime.
Yield Curve normal (+52bps), aber Credit Spreads weiten sich.
PORTFOLIO-CHECK
ASML: Stärkster Gewinner (+63% total), aber underperformed am Rally-Tag.
Stop-Loss bei 1050 EUR empfohlen vor Earnings am 15.4.
Gold ETC: -1.8% in EUR trotz +3.4% Goldpreis in USD — der EUR/USD-Effekt.
EMPFEHLUNGEN
Keine neuen Trades. 14 von 17 Positionen reporten in den nächsten 6 Wochen.
Cash-Quote von 31% ist im Extreme-Fear-Umfeld eine Stärke. Abwarten.
The web dashboard is always there. Everything else is optional — mix as you like:
| Channel | What you get | What you need |
|---|---|---|
| Web dashboard | Everything: portfolio, briefings, recommendations, chat, settings. Add to home screen on your phone. | Nothing extra |
| Browser push | Notifications from the web app (briefings, trades, chat replies, errors) | HTTPS or localhost; VAPID keys (the setup wizard generates them) |
| Telegram | Briefings as messages, chat and trade logging from Telegram | A bot from @BotFather |
| iOS app | Native app for your server, works offline with the last state | Mac with Xcode. Free Apple account (reinstall every 7 days) or paid Developer Program (TestFlight) — see ios/README.md |
| iOS push | Notifications on your iPhone | Paid Apple Developer Program + APNs key |
The setup wizard asks which channels you want. Which events notify you (briefings, trades, chat replies, system errors, …) can be switched per category on the web settings page or in the iOS app. If a system error can't be delivered as a notification, Velora falls back to Telegram when it's configured.
git clone https://github.com/Kingler16/velora.git
cd velora
python3 setup.py # Interactive setup wizard (creates venv + config)
./venv/bin/python -m src.main briefing # First AI briefing
./venv/bin/python -m src.main web # Web dashboard → http://localhost:8080
./venv/bin/python -m src.main bot # Telegram bot (only if you set it up)
The setup wizard guides you through: language, country & tax regime, notification channels, API keys, briefing schedule, and portfolio import (CSV or manual).
Your data stays on your machine: config/settings.json, config/portfolio.json, memory/ and config/keys/ are gitignored.
npm install -g @anthropic-ai/claude-code (requires Claude Max or Pro)Built with FastAPI + Jinja2 + HTMX + Chart.js + TradingView Lightweight Charts. No Node.js needed.
| Page | What It Shows |
|---|---|
| Dashboard | KPI cards, allocation donut, portfolio history chart, sortable holdings table, index ticker bar |
| Portfolio | Per-account breakdown, currency/sector charts, bank accounts, trade logging modal |
| Portfolio › Performance / Tax | Benchmark comparison, monthly returns heatmap, tax-loss harvesting, top/bottom performers |
| Market | 9 indices with descriptions, US/EU macro data, Fear & Greed gauge, earnings calendar |
| Briefings | Market regime, key insights, expandable briefing history with full text |
| Recommendations | Active BUY/SELL cards, WATCH/HOLD watchlist, one-click close, hit-rate KPIs |
| Strategy | Optional investment mandate (target allocation, hard rules) and drift against it. With policy.cash.max set, cash above the band is deployed in fixed monthly tranches (policy.deployment, see config/mandate.example.json) |
| Chat | Conversation with Velora about your portfolio; write actions need your confirmation |
| Settings | API keys, language (de/en), briefing schedule, notifications per category |
Start with python -m src.main web and open http://localhost:8080.
| Input | Action |
|---|---|
/status | Portfolio overview |
/briefing | Trigger AI briefing |
/log | Step-by-step trade form |
AAPL | Deep-dive ticker analysis |
watch RKLB / unwatch RKLB | Manage watchlist |
Bought 10 AAPL @ 180 | Log a trade (natural language, de/en) |
| Any text | Chat with your AI advisor |
Trades logged via Telegram or the web dashboard automatically close matching AI recommendations.
[Cron / Telegram / Web Refresh]
│
▼
main.py ─── data/market.py → yfinance
│ ├── data/macro.py → FRED + ECB + Fear & Greed
│ ├── data/news.py → Brave + Bloomberg + Finnhub
│ └── data/calendar.py → Earnings + events
│
├──→ cache.py → memory/cache/ → Web Dashboard (FastAPI)
│
▼
analysis/claude.py → Claude Code CLI (Opus, high effort)
│
▼
memory.py → briefings, recommendations, theses
│
└──→ delivery → web dashboard / browser push / iOS push (APNs) / Telegram
Velora does not call the Claude API directly. It shells out to Claude Code CLI in --print mode — no API key needed, uses your Max/Pro subscription, Opus model with high effort, prompts via stdin.
python -m src.main briefing # Bi-weekly briefing (for cron)
python -m src.main monthly # Monthly report (for cron)
python -m src.main analyze --ticker AAPL # On-demand analysis
python -m src.main bot # Telegram bot (long-running, optional)
python -m src.main web # Web dashboard (long-running)
python -m src.main web --port 3000 # Custom port
All settings configurable via the web dashboard Settings page or config/settings.json:
| Key | Description |
|---|---|
telegram.bot_token | Bot token from @BotFather (optional) |
telegram.chat_id | Your Telegram chat ID (optional) |
web_push.* | VAPID keys for browser push (python scripts/generate_vapid.py [email protected]) |
apns.* | iOS push: key_path, key_id, team_id, bundle_id — see ios/README.md |
app_store_connect.* | Optional API key so ios/testflight.sh waits for build processing |
web.auth_token | Optional password for dashboard and app (recommended if reachable beyond your home network) |
brave_search.api_key | Brave Search API key |
fred.api_key | FRED API key |
finnhub.api_key | Finnhub API key |
schedule.briefing_days | e.g. ["monday", "thursday"] |
schedule.briefing_time | e.g. "07:00" |
user.language | "de" or "en" |
web.port | Dashboard port (default: 8080) |
Designed for always-on low-power devices like a Raspberry Pi (~90MB RAM), but any Linux box with systemd works.
# On the server
git clone https://github.com/Kingler16/velora.git ~/velora && cd ~/velora
python3 setup.py # configure (channels, keys, portfolio)
bash scripts/setup_server.sh # services + cron
setup_server.sh sets up the web dashboard service, the Telegram bot service (only if Telegram is configured), a daily check that the Claude login still works, and cron jobs for briefings (schedule from settings.json) and the monthly report. Re-run it after changing channels.
Log Claude in once with a long-lived token: claude setup-token, then ./venv/bin/python scripts/set_oauth_token.py.
Updates: git pull, then sudo systemctl restart velora-web velora-bot.
For access on the go, a VPN like Tailscale is the simplest option — the iOS app switches between home Wi-Fi and Tailscale automatically.
| Service | Cost | Notes |
|---|---|---|
| Claude Code CLI | Included in Max ($100/mo) or Pro ($20/mo) | No separate API costs |
| yfinance | Free | No API key needed |
| Brave Search | Free tier: 2,000/month | Sufficient for 2 briefings/week |
| FRED / ECB | Free | US + EU macro data |
| Finnhub | Free tier: 60 calls/min | Sentiment + news |
For educational and informational purposes only. Not financial advice. AI analysis may contain errors. Always do your own research. No liability for financial losses.
MIT