Removed (all recoverable from git history):
- llm_provider_switch_plan.txt: 2026-08-08 DeepSeek/Anthropic multi-provider plan that
was never implemented (llm_client.py is DeepSeek-only, no `anthropic` anywhere)
- docs/superpowers/specs/2026-07-21-centralized-config-design.md: spec of the
config.json migration that merged the same day; its plan sibling was already
deleted in b936d44. app_config's docstring pointer goes with it.
- scripts/compare_oanda_tiingo.py: Phase 0 recent-tail smoke test, superseded by the
Phase 1/2 study scripts (result recorded in the test plan)
- backtest/research/fetch_oanda_m15.py + oanda_td_divergence.py + its by-hour CSV:
the 2026-07-24 Oanda-vs-Twelve-Data study, superseded by the 2026-08-14 feed
divergence report (which keeps a one-line citation)
- backtest/research/burst_profile.py and hour_analysis.py: superseded by burst_map.py
and book_report.py
- backtest/research/panel_payload_drytest_2026-08-15.py: one-off live DeepSeek probe;
its .jsonl evidence stays
Added to git: the three 2026-08-14 research reports (feed divergence, re-pin
verification, signal parity), their CSV, and docs/data_source_change_test_plan.md -
CLAUDE.md cites them as standing evidence and their driver scripts were already
committed. The test plan no longer refers to auto-execution.
Hygiene:
- .gitignore: drop oanda_state.json (daemon removed 2026-09-03) and the vanished
backtest/trailgrid dirs; ignore .coverage
- docs/architecture_diagram.py only writes the PNG when run as a script (it used
to plot on import); test box refreshed to 461 tests / 31 modules; PNG regenerated
(the committed one still drew the removed daemon)
- backtest/instructions_for_backtest.txt: manual-journal head replaced by the
actual reporting checklist; the 2026-09-06 addendum is unchanged
- doc drift: README/publisher list all six published files; news_summary.php says
4 h (not 24 h) and drops the pre-id sha1 fallback whose window closed;
analyze_signal_log lists six lenses and --scan-outcomes; agent_panel cost model
says 4 requests by default; tiingo comment cites the NY-anchored weekend;
Module 21 title no longer says "execution lifecycle"
Suite: 461 passed, 0 failed.
Claude-Session: https://claude.ai/code/session_0156YhogLytzj9P2iH1skGbL
Add panel_verdict, panel_confidence, panel_rationale, panel_takes_json
columns to the signals table (additive, idempotent ALTER TABLE migration
for existing DBs). _report_row now reads the panel fields attach_panel
writes onto the SignalReport, so LLM panel verdicts stop evaporating on
process exit. Also documents the Oanda re-pin (round_trip_cost stays
$0.30 for cross-book comparability and the live TP gate; $0.65 is the
measured headline) and adds the feed-divergence / parity / shadow-scan
study tooling referenced by CLAUDE.md.
The practice API rejects ~15-20 requests/day with a spurious 401
"Insufficient authorization" while the same token succeeds on adjacent
requests; each one silently cost a full scan cycle on that timeframe.
One retry after RETRY_DELAY_S (2 s) absorbs these plus transient
5xx/network faults. Real credential failures fail both attempts and
raise as before; other 4xx (bad request, 429) are never retried.
Tests: transient-401 recovery, persistent-401 raise, 400 not retried.
Claude-Session: https://claude.ai/code/session_01Tujtmmhxj9mxKjPFSoVjmP
Owner decisions from the 2026-08-04 audit report:
- spread-in-exit-triggers: DECLINED (books stay comparable; ledger in
CLAUDE.md records the measured ~4-6% optimism).
- weekend clock: FIXED. The market week is Fri 17:00 -> Sun 18:00 New
York time, not Fri/Sun 22:00 UTC — during US DST the Oanda feed stops
at 21:00 UTC (measured: 159/162 summer weekends). All three weekend
implementations (providers/base, signal_aggregator, gpu_backtest
calendar filter) now share the NY-anchored rule; a new test sweeps
them hourly across both DST transitions and asserts agreement. No NTP
needed — the host clock was fine, the constant was wrong.
Ends the weekly false-"STALE DATA / feed lagging" hours (summer Fri
21-22 UTC, winter Sun 22-23 UTC).
- FB2_EMA_LO (delegated): kept as a research knob but crash-proofed —
an unparseable value degrades to the tuned default 0.5 with a warning
instead of killing every scan. Documented in CLAUDE.md.
Book gates: M15 12mo pinned book byte-identical (+78.7111R, 397
trades). M5 12mo drops exactly ONE trade (556 -> 555, +97.56 ->
+98.64R): a phantom whose signal fired on the final summer-Friday bar
(decision 21:00 UTC = NY 17:00 market close) and was "filled" at
Sunday's reopen 49 hours later, -1.08R — an entry live trading can
never take. The backtest no longer contains it; live == backtest
parity improved. M15 unaffected because hour 21 is in its blocked set.
Suite: 341/341 green.
Claude-Session: https://claude.ai/code/session_01UnZkzeFcTu9agFg8BDh2tn
2026-08-04 Fable 5 QA audit. Three batches sharing one test file, all
verified by runnable repros before fixing, all gated below.
Batch C — telegram alert delivery (verified by h4 harness):
- Failed breach rescan shrinks the cooldown to 30 s instead of
silencing the symbol for 5 min; breach during cooldown is PARKED and
fired on the first tick after expiry (was dropped forever).
- Level cache keyed symbol→TF (M5 scans no longer wipe H1/H4 swing
levels within minutes); breach labels TF-qualified.
- Markdown alert sends get a plain-text fallback (mirroring _safe_edit);
fingerprint commits only when a message actually landed.
- Per-key asyncio.Lock closes the dedup TOCTOU between auto-scan and
breach sends (forced-interleave repro produced duplicate alerts).
- _start_scheduler respawns dead per-TF loops (was all-or-nothing).
- Clean WS close now backs off (was a zero-delay reconnect hot loop);
backoff resets only after a connection that received ticks; TD
iter_ticks fires on_connect on the first RECEIVED frame (invalid key
can no longer pin the backoff at minimum).
- Scan-failure logs carry the real traceback (was "NoneType: None").
Batch E — providers (verified by h5 harness):
- fetch_range: 3 consecutive chunk ERRORS raise instead of returning
silently truncated history (verified 731-day silent gap repro; also
covers 429 storms); pagination no-progress guard; head-truncation
ERROR log.
- Single-flight per (symbol,TF): 4 concurrent scans = 1 API call (was
4); cache records requested depth so a longer-lookback call refetches.
- Multi-bar forming/future tail fully dropped (was iloc[:-1] once).
- Oanda: malformed candles skipped, never 0.0-defaulted; cleaning drops
non-positive-price rows BEFORE OHLC repair; inf returns neutralized so
one corrupt bar can no longer disarm the whole 10-sigma filter.
Batch F — backtester internals (verified by h6 harness):
- NaN bar in a trailing window exits at the last valid bar (was NaN R
poisoning every metric); static-path TIMEOUT NaN guard;
compute_metrics excludes non-finite R loudly. Zero real-data exposure
(0 NaN rows in all 10 Oanda parquets) — defensive.
- fetch_history ERROR-logs pre-2026-07-08 cache files (cleaned-bars era)
when loaded for raw-path resolution.
Gates: suite 333/333 green; pinned M15 12mo book reproduced identically
(397 trades, +78.7111R) after E and after F; M5 HEAD-config baseline
reproduced identically (556 trades, +97.5604R) after F.
Claude-Session: https://claude.ai/code/session_01UnZkzeFcTu9agFg8BDh2tn
The monitor has been dead since the Oanda feed switch (Oanda's provider
does not stream). New config.json scalar data_providers.streaming selects
a tick-stream provider independent of the candle feed; telegram_bot
resolves it via _get_streaming_dp() (separate instance, candle cache
stays on the active provider) and falls back to the active provider when
unset — behavior is byte-identical to before when the key is absent.
TiingoDataProvider grows the two sanctioned streaming overrides:
supports_streaming=True and iter_ticks() on wss://api.tiingo.com/fx
(mid price only — the no-spread rule extends to the stream). Two
deliberate divergences from the TD reference implementation, both
because Tiingo authenticates inside the socket: server "E" messages
raise TiingoDataError (a swallowed auth error would end the generator
cleanly and reconnect-hammer with zero backoff), and on_connect fires on
the first received message rather than after the subscribe send. Field
order of quote frames was live-verified: ask is data[7], not data[6] as
the docs table suggests (data[6] is askSize).
make_data_provider(config, name=None) now takes an explicit provider
name; keyless streaming providers disable the monitor with one warning
instead of a traceback-per-minute loop.
Tests 266 → 273 (WS parser, ticker round-trip, streaming config scalar,
explicit-name dispatch, streaming-dp resolution, monitor auto-disable).
Claude-Session: https://claude.ai/code/session_01Nsd8HcrV67GbCZXxZEn1Ma
One class per provider (TwelveData, Oanda, Tiingo), all subclassing
BaseDataProvider which owns cache, validation/cleaning, freshness +
boundary guarantees, fetch_range pagination, and streaming stubs —
subclasses implement only _fetch_raw. config.json data_providers.active
selects the live feed; data_provider.py / oanda_provider.py stay as
back-compat shims. Module 16 provider-abstraction tests (253 total green).
Claude-Session: https://claude.ai/code/session_01Mc7XKYyPySVXiUpcN82QEK
Harwinder Sekhon