Commit Graph

88 Commits

Author SHA1 Message Date
Celes Renata 1a5fb2e36a fix: recovery sweep touches updated_at after re-enqueue to prevent duplicate flooding every 5 min 2026-04-16 19:13:10 +00:00
Celes Renata 693d9e0d60 fix: reduce LLM timeouts — truncate docs to 8k/6k chars, cut num_predict 16k→4k, tighten prompts, trim anti-hallucination rules 2026-04-16 18:56:11 +00:00
Celes Renata 60cfb7618e fix: recovery sweep skips docs that already have global_events — prevents re-enqueue loop 2026-04-16 18:27:21 +00:00
Celes Renata 1043710b6d fix: track last_published_at per source to avoid re-fetching same articles — applies to both news_api and macro_news 2026-04-16 18:12:12 +00:00
Celes Renata 513310abba fix: stop tagging all macro_news articles as macro_event — default to article, let extractor reclassify. Also reduced fetch limit to 20 and cadence to 30min 2026-04-16 18:09:50 +00:00
Celes Renata f83577480f fix: alternate extractor between macro and extraction queues (1:2 ratio) to prevent starvation 2026-04-16 17:45:25 +00:00
Celes Renata 2440cddd37 feat: add comprehensive table retention cleanup to scheduler — 10 tables with per-table retention windows 2026-04-16 15:56:50 +00:00
Celes Renata 63287903d0 feat: wire up stop levels, circuit breaker daily loss, profit-taking, real portfolio/decisions/history endpoints 2026-04-16 15:52:46 +00:00
Celes Renata 1329df0bbf feat: sell execution, correlation matrix from market data, US market holiday awareness
- Sell path: looks up existing position, sells full quantity, returns proceeds to pool
- Correlation matrix: computed from 30-day market_snapshots on startup + every 5min
- Holidays: 10 major US market holidays for 2026 checked in trading window functions
2026-04-16 15:36:49 +00:00
Celes Renata 2e77cf32fd fix: critical — track capital properly: load invested positions on startup, deduct on act, sync every 5min 2026-04-16 15:29:28 +00:00
Celes Renata 9a8d36068a fix: convert Decimal to float in API responses instead of string — fixes positions page crash 2026-04-16 15:25:40 +00:00
Celes Renata 354c3d484a fix: fetch current prices from market_snapshots before evaluating recommendations — fixes 'Invalid current price' skip 2026-04-16 15:17:49 +00:00
Celes Renata 2a6aac47a6 fix: add decision logging to trading engine, flushed 103k stale dedup keys 2026-04-16 15:12:58 +00:00
Celes Renata c114e77b1c fix: limit recommendation poll to 50 per cycle to prevent 85k-rec processing stall, add poll logging 2026-04-16 15:05:26 +00:00
Celes Renata 136b149000 fix: add logging config to trading engine, add /api/trading/debug diagnostic endpoint 2026-04-16 14:55:42 +00:00
Celes Renata 6bab199159 fix: trend_windows now upserts instead of accumulating (7.5GB→4MB), add competitive signal retention cleanup 2026-04-16 14:32:24 +00:00
Celes Renata 58a8726306 feat: add paper trading capital controls — API endpoint + UI with presets, fix status/metrics to read real state, fix migration duplicates 2026-04-16 14:06:30 +00:00
Celes Renata 14e411daf9 fix: trading status and metrics endpoints now read real portfolio state instead of hardcoded zeros 2026-04-16 14:02:38 +00:00
Celes Renata 5cc64498c0 fix: skip already-propagated docs in aggregation, limit to last hour + 10 docs instead of 50 2026-04-16 09:35:11 +00:00
Celes Renata f0887afd9b fix: cap competitive signals to 500 most recent per window to prevent 67k row aggregation bottleneck 2026-04-16 09:24:31 +00:00
Celes Renata 58a05ca322 feat: add SSE stream for live pipeline status, add all 10 queues + DLQs, configure nginx for SSE 2026-04-16 08:15:44 +00:00
Celes Renata 87579d68da fix: add stale document recovery sweep to scheduler, re-enqueues orphaned parsed docs every 5 min 2026-04-16 07:59:30 +00:00
Celes Renata cdc825619e feat: add live queue depths to pipeline health API and dashboard 2026-04-16 07:49:07 +00:00
Celes Renata 0ee7f26633 feat: raise market_api rate to 20/min, add global Polygon cap at 45/min, add rate-limit API + watchlist warning 2026-04-16 07:26:10 +00:00
Celes Renata 1107d34027 fix: SQL Explorer handles comments and shows descriptive errors
- Strip SQL comments (-- and /* */) before checking for SELECT,
  so queries with leading comments don't get rejected
- Show the actual error detail from the API response instead of
  generic 'API error 400' in the SQL Explorer UI
2026-04-16 05:25:45 +00:00
Celes Renata c4666c071b feat: wire Gmail SMTP notifications with app password
Replaced the Gmail API (OAuth2) notification delivery with plain
SMTP using a Gmail app password. Much simpler setup — no Google
Cloud project, no OAuth2 flow, no extra dependencies.

- Rewrote _send_gmail() to use smtplib with smtp.gmail.com:587 TLS
- Added stonks-gmail-secrets to Helm chart (GMAIL_SENDER,
  GMAIL_RECIPIENT, GMAIL_APP_PASSWORD)
- Added gmail secret to trading-engine deployment
- Updated runmefirst.sh to read gmail.app from kube dir
- Sender/recipient: celes@celestium.life
2026-04-16 02:37:40 +00:00
Celes Renata 88c9f50371 fix: treat 404 on Alpaca positions endpoint as empty result
Alpaca returns 404 when you don't hold a position in a ticker.
The ingestion worker was logging this as an error and incrementing
the failure count. Now returns an empty items list instead, since
'no position' is a valid state, not an error.
2026-04-16 01:33:35 +00:00
Celes Renata 949324dc89 feat: SQL Explorer with PostgreSQL schema browser and pre-built queries
The SQL Explorer was querying Trino which has zero tables. Rewrote to
use PostgreSQL directly:

Backend:
- GET /api/analytics/pg-schema: returns all public tables with column
  names, types, and nullability from information_schema
- POST /api/analytics/pg-query: read-only SQL execution against
  PostgreSQL with SELECT-only enforcement, auto LIMIT, and descriptive
  error messages for syntax/table/query errors

Frontend:
- Schema browser shows all PostgreSQL tables with columns and types
- Click a table name → generates SELECT * FROM table LIMIT 100
- Pre-built Queries section with 12 seeded queries covering companies,
  recommendations, trends, market prices, documents, global events,
  trading decisions, ingestion health, reserve pool, sector exposure
- User-saved queries shown separately with delete buttons
- Chart builder, Monaco editor, and save functionality preserved

Migration 021: seeds 12 pre-built saved queries
2026-04-16 01:06:49 +00:00
Celes Renata 6eda988e3b fix: strip /v2 suffix from broker base URL to prevent doubled path
The alpaca.url config file contains https://paper-api.alpaca.markets/v2
but the adapter code also prepends /v2/ to all paths, resulting in
/v2/v2/positions which returns 404. Now strips trailing /v2 or /v1
from the configured base URL since the adapter manages API versioning.

This was causing 1,017 consecutive broker sync failures.
2026-04-16 00:45:19 +00:00
Celes Renata 88c2bc84a1 feat: upgrade paper trading to $100k moderate tier
Paper money has no downside — bigger capital exposes more model
behavior: position sizing, diversification, sector exposure,
correlation checks, circuit breakers, reserve pool siphoning,
and risk tier auto-adjustment all become meaningful.

- risk_tier: conservative → moderate (min_confidence 0.55)
- absolute_position_cap: $25 → $10,000
- max_open_positions: 5 → 10
- initial portfolio value: $500 → $100,000
- Updated migration 019, Helm values, and engine default
2026-04-16 00:37:35 +00:00
Celes Renata d21110b3d2 fix: backtest skips duplicate ticker positions to prevent overwrite
When multiple recommendations for the same ticker produce 'act'
decisions, the second one would overwrite the first in
simulated_positions, losing the first position's value and causing
incorrect portfolio value calculations. Now skips if already holding.
2026-04-16 00:28:59 +00:00
Celes Renata 2eaf6dc025 fix: upgrade backtest day summary to WARNING level for visibility 2026-04-16 00:21:46 +00:00
Celes Renata bad7e02e53 fix: add first-skip-reason logging to backtest replay for debugging 2026-04-16 00:16:32 +00:00
Celes Renata 4634f1f3fc fix: remove unused skip_count variable (lint) 2026-04-16 00:07:29 +00:00
Celes Renata ff5055ee4e fix: backtest replay field mapping and logging
- Map DB 'id' field to 'recommendation_id' for evaluate_recommendation()
- Ensure confidence is cast to float (asyncpg may return Decimal)
- Add per-day logging showing rec count, act/skip, positions, pool balance
- Helps diagnose why backtests produce 0 trades
2026-04-15 22:55:26 +00:00
Celes Renata 4501bbebd4 feat: add Polygon grouped daily endpoint for broad market data
Two tiers of market data:
1. Per-ticker prev bars (existing 50 sources, 15-min cadence) for
   watchlist detail — trading decisions, stop-loss, position sizing
2. Grouped daily (new single source, once per day) for broad market
   context — correlation analysis, sector rotation, competitive intel

Changes:
- Add grouped_daily endpoint to PolygonMarketAdapter with auto date
  calculation (previous trading day, skip weekends)
- Add fetch_global_market_sources() to scheduler for sources without
  company_id, scheduled once daily (86400s cadence)
- Update _persist_market_items to use item-level ticker from T field
  and look up company_id dynamically for grouped daily bars
- Migration 020: make company_id nullable on sources and
  market_snapshots tables, add grouped daily source row
- Fix backtest replay to query market_snapshots data->>'c' for prices
2026-04-15 22:38:18 +00:00
Celes Renata ea6c2b3f54 fix: market data rate limiting and backtest price lookup
- Increase market_api polling cadence from 60s to 900s (15 min).
  The prev-day bar endpoint returns the same data all day, so polling
  every minute wastes API quota. 50 tickers at 15-min cadence = ~3.3
  req/min, well within the 5/min rate limit.
- Reduce market_api rate limit from 30/min to 5/min to match.
- Fix backtest replay to query market_snapshots with data->>'c' for
  close prices instead of nonexistent market_data.close_price column.
- Enrich backtest recommendations with prices from market_snapshots
  and sectors from companies table.
2026-04-15 22:19:44 +00:00
Celes Renata 69eb366bf9 fix: backtest simulation uses ET timezone for trading window
The simulated timestamp was 10:00 UTC (6:00 AM ET) which is outside
the trading window. Changed to 11:00 AM ET so backtested decisions
actually pass the trading window check.
2026-04-15 21:48:32 +00:00
Celes Renata 70bad7709a feat: wire live decision loop and enable paper trading
Phase 2 of the autonomous trading engine:

- Replace start()/stop() stubs with real async implementations
- Decision loop: polls recommendations from PostgreSQL, deduplicates
  via Redis, evaluates through the full pipeline, submits orders to
  stonks:queue:broker_orders
- Stop-loss monitor: fetches prices from Polygon API, checks crossings,
  submits immediate sell orders, safety sell after 15 min without data
- Performance loop: computes metrics every 5 min during market hours,
  persists daily snapshots at market close
- Risk tier scheduler: evaluates daily at 16:00 ET, persists tier changes
- Rebalance scheduler: evaluates Monday 09:45 ET, respects circuit breaker
- Notification dispatch: SNS + Gmail with rate limiting and retry
- Backtest replay: fetches historical data, simulates decisions, persists
- Real asyncpg/redis connections in FastAPI lifespan (graceful degradation)
- Migration 019: enable paper trading with conservative tier, 5 cap
- Added max_open_positions to TradingConfig with env var loading
- Phase 2 tasks added to autonomous-trading-engine spec
2026-04-15 20:52:28 +00:00
Celes Renata 3ff910433f fix: reject empty LLM classifications for global events
When the LLM returns empty summary and no key facts, raise ValueError
so the retry logic kicks in instead of persisting an empty event.
Also strip whitespace from summary and filter empty key_facts entries.

Cleaned up 17 empty events from the database.
2026-04-15 19:46:31 +00:00
Celes Renata 326c409d63 fix: use _global fallback for empty ticker in MinIO storage paths
Macro news documents have no ticker, causing upload_normalized_text
and upload_parser_output to produce paths like parsed//2026/...
which MinIO rejects as XMinioInvalidObjectName. Use '_global' as
the path segment when ticker is empty, matching the existing
macro prefix pattern in upload_raw_document.
2026-04-15 19:25:23 +00:00
Celes Renata 4ffde8cc06 feat: autonomous trading engine — full implementation
- Database migration 018 with 13 tables for trading engine state
- Trading engine service (services/trading/) with 12 pure computation modules:
  position sizer, stop-loss manager, reserve pool, circuit breaker,
  risk tier controller, correlation matrix, tax lots, trading window,
  gradual entry, notifications, micro-trading, backtester
- Core TradingEngine with pre-trade evaluation pipeline and integration wiring
- FastAPI HTTP service with 14 endpoints (health, config, decisions, metrics, backtest)
- Performance tracker with Sharpe ratio, drawdown, profit factor computation
- 194 Python tests (165 property-based + 29 integration)
- Frontend: 13 TanStack Query hooks, 7 dashboard panels, tabbed Trading Engine page
- Helm chart entry, network policy, nginx proxy, ingress for trading-engine
- Shared infrastructure: enums, Redis keys, TradingConfig in AppConfig
2026-04-15 16:12:22 +00:00
Celes Renata da86132f0c fix: add num_predict=16384 to prevent output truncation on large articles 2026-04-15 03:11:13 +00:00
Celes Renata b8a2cdc52a fix: fill default values for missing fields in truncated LLM output 2026-04-15 03:08:10 +00:00
Celes Renata 00044af993 fix: switch to think=false with json-repair — 20x faster extraction 2026-04-15 02:54:39 +00:00
Celes Renata 4f2ae23d42 fix: set num_predict=16384 so model has token budget for thinking + content 2026-04-15 01:47:00 +00:00
Celes Renata 46b069a748 fix: switch to non-streaming Ollama calls — streaming breaks thinking mode 2026-04-15 01:19:17 +00:00
Celes Renata ffe19eb23a fix: handle empty ticker in MinIO storage paths, clean up debug log 2026-04-15 00:39:53 +00:00
Celes Renata 8b5b692d3c fix: update stall timer during thinking phase to prevent premature stream abort 2026-04-15 00:06:49 +00:00
Celes Renata 01726af360 fix: remove think=false (Ollama bug #14645), bump max_tokens to 32k 2026-04-14 23:50:28 +00:00