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Metron — portfolio analytics, measured. Multi-tenant portfolio-analytics SaaS (Nous Ergon). No AI, no ads, no advice, read-only. AGPL-3.0.
翻譯準備中
取自儲存庫 README
正在載入完整 README。 以下為已儲存的摘錄。
Portfolio intelligence for people with real brokerage accounts.
Metron is a multi-tenant dashboard for institutional-grade portfolio analytics on
your real accounts — true returns, attribution, factor risk, scenarios, income, and
tax clarity. An intelligence layer surfaces trends and analyses you'd otherwise miss.
No ads/trackers, read-only. We compute; we never tell you what to trade.
A Nous Ergon product, hosted at metron.nousergon.ai.檢視完整 README ↗Portfolio intelligence for people with real brokerage accounts.
Metron is a multi-tenant dashboard for institutional-grade portfolio analytics on your real accounts — true returns, attribution, factor risk, scenarios, income, and tax clarity. An intelligence layer surfaces trends and analyses you'd otherwise miss. No ads/trackers, read-only. We compute; we never tell you what to trade.
A Nous Ergon product, hosted at metron.nousergon.ai.
Metron is AGPL-3.0. Self-host it freely. The hosted service exists for the zero-ops convenience (managed sync, hosting, updates) and the proprietary data feeds that don't ship in this repo — that's the commercial side; the product itself is open.
The quant core (factor risk, attribution, returns, VaR/CVaR, riskstats) is not
duplicated here — it lives in the public, MIT-licensed
nousergon-lib and is imported.
One codebase, two top-level Python packages (+ a web frontend):
| Path | What |
|---|---|
portfolio_analytics/ | The pure engine. domain/ (ledger, realized income, tax lots, stress), broker_io/ (IBKR Flex, SnapTrade, transaction tranching), ingestion/ (FDX canonical schema + bronze/silver store + broker connectors + CanonicalReader). No web/cloud coupling; fully unit-tested. |
api/ | FastAPI service + the multi-tenant Postgres schema (api/db/models.py) over the engine. Tenant-isolated via Postgres RLS in prod. |
web/ | Next.js + Tremor frontend — shipped and live. |
Proprietary runtime bits (LLM advisor prompt templates, private signal feeds) are never committed — they load at runtime from gitignored config, so a self-host runs the full open product without them and the hosted product layers them on.
python -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'
pytest # engine + API
uvicorn api.main:app --reload
# → http://127.0.0.1:8000/health · /meta · /docs
DATABASE_URL defaults to a local SQLite file (zero vendor cost). Point it at Postgres
(postgresql+psycopg://…) for production — no model changes required.
Coverage is measured over the whole portfolio_analytics + api source and enforced:
pytest exits non-zero below the floor in .github/workflows/ci.yml
(--cov-fail-under=95), and the badge above renders the figure CI last measured on
main.
Metron is the successor to the personal RoboDashboard (Streamlit), whose front-end-agnostic engine was extracted here on 2026-06-09. RoboDashboard is being retired once Metron reaches feature parity.