feat(pipeline): #503 Phase 3a — warmup_window validation#508
Merged
MDUYN merged 4 commits intoMay 4, 2026
Merged
Conversation
…ion (#503 phase 3a) When a strategy declares pipelines=[...], TradingStrategy.__init__ now checks that every OHLCV data source has warmup_window >= pipeline.required_window() and raises OperationalException with an actionable message otherwise. Without this guard a live bot silently trades on NaN columns until enough bars accrue. - 6 new tests covering: sufficient/insufficient/unset warmup, no OHLCV source, no pipelines (zero-cost), and multi-pipeline max-window aggregation. - pipelines-live.md: mark warmup validation shipped.
…esilience (#503 phase 3b/3c/3d) Completes the framework-side hardening for live pipeline trading. Async/batched CCXT fetch (#503's last bullet) is deferred to a follow-up because it needs live integration testing. Phase 3b — Live envelope validation * Strategies that declare pipelines and run in any non-BACKTEST environment are validated once per run against a v1 envelope: - Maximum 50 unique OHLCV symbols per strategy - Daily-or-coarser timeframes only * Violations raise OperationalException with an actionable message naming the strategy and the offending symbols / timeframe. * Backtests are completely unaffected (no-op). Phase 3c — Universe-refresh cadence * New Pipeline class attribute 'refresh_universe_every: timedelta'. * When set, the engine caches the surviving symbol set after each refresh and reuses it on subsequent bars within the cadence, skipping universe-filter evaluation entirely. Factors are still recomputed every bar. * Default None preserves Phase 1 / Phase 2 behaviour exactly. Phase 3d — Per-pipeline resilience * In live (non-BACKTEST) mode a single failing pipeline.evaluate() is logged and the iteration continues with an empty output for that pipeline. Backtests still re-raise so failures stay deterministic. Tests * 12 new tests in tests/app/test_pipeline_live_hardening.py covering sub-daily rejection, >50-symbol rejection, no-pipeline pass-through, backtest skip, one-shot validation, all four cadence-cache cases, and live-vs-backtest error handling. Docs * docusaurus/docs/Advanced Concepts/pipelines-live.md updated with Live envelope, Universe refresh cadence, and Per-pipeline resilience sections. Full regression: 1344 passed, 3 skipped.
CI uses 'unittest discover' and does not install pytest as a dev dependency, so 'import pytest' caused ImportError on the runner. Rewrites all assertions/raises to native unittest.TestCase APIs. No production code changes; behavior covered is identical.
Same fix as the previous commit — CI uses 'unittest discover' which doesn't have pytest available. Converts test_vector_pipeline_engine, test_factor_arithmetic_and_lazy, and test_vector_backtest_pipeline_injection to native unittest.TestCase, with pytest.approx -> assertAlmostEqual, pytest.raises -> assertRaises, parametrize -> subTest, monkeypatch -> unittest.mock.patch.object, and caplog -> assertLogs.
This was referenced May 4, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #507 (Phase 2). Closes the first milestone of #503.
What
When a strategy declares
pipelines = [...],TradingStrategy.__init__now validates that every OHLCV data source haswarmup_window >= pipeline.required_window(). If any source is short — orwarmup_windowis left unset — strategy instantiation raisesOperationalExceptionwith an actionable message listing the offending sources and the required window size.Why
Without this guard, a live bot silently trades on NaN factor columns until enough bars accrue. This is a known footgun for live deployments on AWS Lambda / Azure Functions where warm-container restarts can swallow startup warnings.
What's not in scope
These will follow in stacked PRs.
Tests
Docs
Refs: #503