Upgrade BTC oracle to Fire Horse v10 architecture

[jpbpari-sys]

Motivation

• Replace the previous simple polling/threaded oracle with a single-file, production-oriented async architecture that supports live websocket ingestion, richer spectral/evolution engines, risk-aware sizing, telemetry and airdrop accounting.
• Provide a plug‑and‑play path for optional live trading via CCXT while keeping a realistic position simulator and persistent logging for reproducible backtesting and observability.

Description

• Reworked btc_agentic_oracle.py into the “Fire Horse v10” architecture by adding an async Binance websocket pipeline that produces dual-timeframe returns and feeds the spectral ridge/ridge-swarm engine and evolutionary agents; the core spectral and ridge logic remains but is integrated into the async flow.
• Added portfolio and position logic including PortfolioManager, a position simulator (fees/slippage/drawdown), live-trade hooks using ccxt.async_support and simulate_trade for paper mode.
• Added SQLite-backed telemetry with init_db, log_trade, log_metrics, and log_airdrop and new FastAPI endpoints GET /api/trades, GET /api/metrics, and GET /api/airdrops for the dashboard.
• Upgraded the web UI served at / to a richer dashboard (Chart.js charts, websocket updates, airdrop display) and changed the boot path to call init_db() and start fetch_live_tick() before running the server.

Testing

• Ran python -m py_compile btc_agentic_oracle.py and it succeeded with no syntax errors.
• Attempted to run python btc_agentic_oracle.py to start the full app but it failed in this environment with ModuleNotFoundError: No module named 'aiohttp', so the live async stack could not be launched here; install aiohttp (and other runtime deps) to run the server end-to-end.
• Basic static inspection and lightweight runtime checks (imports/structure) performed during development; no further automated integration tests were executed in this environment.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9377d7c539

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Gate live orders on position state before create_order

In live mode this branch submits execute_real_trade(sig, ...) on every cooldown cycle without checking whether the current position already matches sig, unlike simulate_trade which suppresses duplicate same-side entries. With a persistent LONG/SHORT signal, this will place repeated market orders every 5 seconds and unintentionally stack exposure instead of holding or flipping once.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Classify volatility regime before normalizing returns

returns_fast is z-score normalized here, but regime detection later uses np.std(signal) thresholds to decide LOW/MID/HIGH volatility. After this normalization, the standard deviation is almost always ~1, so the strategy will stay in MID_VOL nearly all the time and the low/high-volatility control paths (confidence scaling and signal gating) are effectively disabled.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Run tick stream task on Uvicorn's event loop

The tick ingestor is scheduled on the loop returned by asyncio.get_event_loop() immediately before uvicorn.run(...); on current Uvicorn versions, run starts the server with its own event loop, so this pre-created task is not driven. In that runtime configuration, market websocket ingestion never starts and the oracle pipeline stays idle.

Useful? React with 👍 / 👎.