jgrusewski 54e3d50db8 backtest: fusion (regime-adaptive QQQ<->crisis-alpha) FAILS OOS, keep static GD ...

Fused all ideas: QQQ core + foxhunt controllers as a ROTATOR (edge-decay-trust + resurrection +
drawdown-CB) rotating QQQ<->crisis-alpha by regime. Pre-registered criterion: beat static GD on OOS
Calmar. RESULT: full-sample fusion Calmar 0.41 > GD 0.34 (IS-flattered by 2008), but OOS (2018-2026)
static GD 0.64 > fusion 0.58, even pure QQQ 0.59 > fusion -- GD strictly dominates OOS (higher CAGR +
lower DD). Regime-rotation whipsaw + late-re-entry cost exceeds the protection benefit in bulls.
Verdict: fusion fails OOS -> keep the simpler static GD. Simple beats fancy; only OOS separates them.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

2026-06-08 23:01:04 +02:00

..

audit-manifest

audit(rust-consts): catch literal-vs-const drift + cleanup BOOK_LEVELS=10

2026-05-24 17:39:40 +02:00

surfer

backtest: fusion (regime-adaptive QQQ<->crisis-alpha) FAILS OOS, keep static GD

2026-06-08 23:01:04 +02:00

tests

plan5(task5-B): pivot multi-seed Argo from N×K (seed,fold) to N seed-only fanout

2026-04-26 14:11:32 +02:00

validation

plan5(task4): tiered-exit validation script suite (tier1/2/3 checks)

2026-04-26 12:33:28 +02:00

aggregate-multi-seed-metrics.py

plan5(task5-A): wire tier 2/3 val_* metrics into HEALTH_DIAG

2026-04-26 13:04:39 +02:00

aggregate-norm-stats.py

plan5(task1B): metric aggregation across multi-seed runs

2026-04-26 10:58:42 +02:00

alpha_pipeline.sh

refactor(alpha_baseline): rename, drop conditionals, strip dead paths

2026-05-16 01:02:44 +02:00

alpha-rl-run.sh

feat(scripts): alpha-rl-run.sh production wrapper + session memory updates

2026-05-26 11:39:53 +02:00

argo-alpha-cv.sh

feat(alpha): decision-stride + cluster 9-fold CV workflow

2026-05-16 18:04:34 +02:00

argo-alpha-perception.sh

revert(loader): multi-resolution default '1:32' (single-scale) after htpp6 falsification

2026-05-22 21:21:21 +02:00

argo-alpha-rl.sh

feat(infra): plumb FOXHUNT_BAND_ENABLED through alpha-rl template

2026-06-04 09:05:25 +02:00

argo-compile-deploy.sh

…

argo-download.sh

…

argo-lob-sweep.sh

feat(argo): run-sweep template + batched-mode in argo-lob-sweep.sh

2026-05-19 18:15:22 +02:00

argo-nsys.sh

fix(argo): nsys upload to foxhunt-training-results bucket (real bucket name)

2026-04-28 22:56:31 +02:00

argo-precompute.sh

fix: argo-precompute.sh uses train workflow + regenerated v2 fxcache

2026-04-17 01:29:42 +02:00

argo-sanitizer.sh

infra(argo): add sanitizer-test + nsys-test workflow templates for L40S smoke validation

2026-04-28 21:57:32 +02:00

argo-smoke.sh

infra(argo): smoke-test --clean-cache parameter for build-cache-isolated bisect

2026-04-29 11:04:18 +02:00

argo-test.sh

infra(argo): default GPU pool to ci-training-l40s (sm_89) per feedback_default_to_l40s_pool

2026-05-14 14:22:19 +02:00

argo-train.sh

fix(argo): add alpha-rl to argo-train.sh model validation

2026-05-25 09:26:19 +02:00

audit_sp18_consumers.sh

feat(sp18 v2 P1.T1+T2): pre-dispatch consumer-audit script + pre-commit hook

2026-05-09 02:23:57 +02:00

audit-diag.sh

docs(sp20): trader-grade trade management spec + audit infrastructure

2026-05-24 16:28:39 +02:00

audit-isv.sh

fix(rl): wire TrailTighten/TrailLoosen + SP20 P1+P5 foundation

2026-05-24 16:47:53 +02:00

audit-rust-consts.sh

audit(rust-consts): catch literal-vs-const drift + cleanup BOOK_LEVELS=10

2026-05-24 17:39:40 +02:00

audit-wiring.sh

fix(rl): wire TrailTighten/TrailLoosen + SP20 P1+P5 foundation

2026-05-24 16:47:53 +02:00

build_2q_fxcache.sh

scripts(alpha): rename + add cross-quarter validation pipeline

2026-05-16 00:42:07 +02:00

build_docker_images.sh

…

check-gpu-availability.sh

…

compare-backward-state.py

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

compare-mamba2-state.py

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

compare-nsys-profiles.py

plan5(task3): A.4.1 nsys profile harness with regression-comparison script

2026-04-26 12:25:35 +02:00

compare-per-state.py

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

compare-rl-state.py

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

cuda-kernel-guard.sh

…

deploy-secrets.sh

…

determinism-check.sh

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

devpod-setup.sh

…

docker-init-migrations.sh

…

enforce_coverage.sh

…

entrypoint-generic.sh

…

gather-multi-seed-metrics.sh

plan5(task1B): metric aggregation across multi-seed runs

2026-04-26 10:58:42 +02:00

generate_dev_certs.sh

…

generate_sweep_variants.py

refactor(per-horizon): N_HORIZONS 5→3 — sweep configs + generator script

2026-05-22 01:40:13 +02:00

generate-training-pipeline.sh

…

gpu-hotpath-guard.sh

…

gpu-hotpath-hook.sh

…

health-check.sh

…

init-db-dev.sql

…

init-db.sql

…

install_torch_gpu.sh

feat(surfer): Phase 0 GPU floor + validation harness — first real (weak) OOS trend edge

2026-06-06 01:31:37 +02:00

local-mid-smoke.sh

feat(ml-alpha): deterministic same-seed training + Tier 1.5 fast-dev-cycle

2026-06-02 17:56:00 +02:00

measure_crossasset.py

feat(scripts): cross-asset trend/lead-lag harness (6E/ES/NQ/ZN OHLCV)

2026-06-05 20:39:31 +02:00

measure_lowfreq_ic.py

feat(scripts): lower-frequency directional IC harness (intraday + daily + backtest)

2026-06-05 19:58:18 +02:00

measure_ofi_ic.py

feat(scripts): OFI information-coefficient signal-validation harness

2026-06-05 19:14:26 +02:00

measure_passive_mm_queue.py

feat(scripts): queue-aware passive-MM fill model (decisive viability test)

2026-06-05 19:28:05 +02:00

measure_passive_mm.py

feat(scripts): passive market-making markout/adverse-selection harness

2026-06-05 19:21:30 +02:00

measure_reward_alignment.py

feat(scripts): reward-pnl alignment measurement harness (RA-G3 gate)

2026-06-05 09:23:29 +02:00

populate-metric-bands-from-runs.py

plan5(task2): A.4 regression-detection hard-stop on 2N consecutive error-band

2026-04-26 11:49:14 +02:00

pre-commit-hook.sh

refactor(rl): delete host-side PER + diag-every band-aid (greenfield prep)

2026-05-26 00:18:11 +02:00

prepare-sqlx-offline.sh

…

ptx-cache-invalidate.sh

…

refresh-test-data.sh

…

requirements.txt

plan5(task1B): metric aggregation across multi-seed runs

2026-04-26 10:58:42 +02:00

safe-node-replace.sh

…

setup-database.sh

…

start-fxt.sh

…

start.sh

…

stop.sh

…

tier1_5_verdict.py

fix(ml-alpha): Phase 0 — per-step authoritative USD pnl in diag + fix mislabeled fields

2026-06-02 19:55:31 +02:00

validate_data_quality.sh

…

walk_forward_cv.sh

refactor(alpha_baseline): rename, drop conditionals, strip dead paths

2026-05-16 01:02:44 +02:00