Menu
CPI/GDP/retail sales date awareness and VIX regime scaling for calendar features
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Normalize long-form CODEX cycle folders to short form before notebooks run. Trigger: cyc001_reg001_*, hard-coded cyc paths breaking, staged CODEX raw data failing in Notebooks 1/2.
v5.6.0 joint multi-TF model: single model per symbol with broadcast 1Hour context replaces dual 15Min/1Hour models. Trigger: (1) replacing weighted-voting model aggregation, (2) adding broadcast features to vectorized env, (3) limited training data + worried about overfitting from doubling obs_dim, (4) backtest builder mismatch with newer feature counts.
DEPRECATED in v5.6.0 — see joint-multi-tf-v560 skill. Documents the v5.2.0 dual-model approach (train separate 15Min/1Hour models, combine via weighted voting). Still relevant for: (1) loading legacy v5.5.0 dual models, (2) understanding the historical aggregation layer, (3) resampling pattern via origin='start'.
Surface a shipped-but-undocumented CLI feature in user-facing docs. Trigger: user reports a known feature missing from README/readthedocs even though the CLI command exists.
KINTSUGI Snakefile + CLI changes that route SLURM jobs around accounts saturated by OTHER users on the same QOS pool. Trigger: QOSGrpMemLimit, jobs stuck pending despite available GPU slots in config, noisy neighbor on shared QOS, multi-user investment pool exhaustion, _build_cycle_assignment static-vs-live.
KINTSUGI SLURM batch processing: Maximize throughput using multi-account resource calculation with GPU+CPU pools per account. Trigger: SLURM job submission, batch processing, resource maximization, GPU+CPU concurrent, headless processing, resource pool.
| name | macro-calendar-modulation |
| description | CPI/GDP/retail sales date awareness and VIX regime scaling for calendar features |
| author | Claude Code |
| date | "2026-02-24T00:00:00.000Z" |
| version | 4.4.0 |
| Item | Details |
|---|---|
| Date | 2026-02-24 |
| Goal | Add CPI/GDP/retail sales awareness and VIX regime scaling to calendar features without changing obs_dim |
| Environment | Python 3.11+, OpenBB FRED provider |
| Status | Success |
Calendar features only knew FOMC + employment first Friday. No VIX level, no CPI/GDP dates. Earnings cause 5-10% average moves; CPI/GDP releases move markets significantly.
Design constraint: NUM_FEATURES must stay at 7 (no obs_dim change). Solution: modulate existing hour_weight value.
CPI_RELEASE_DATES = [datetime(2024,1,11), datetime(2024,2,13), ...] # 2024-2026
GDP_RELEASE_DATES = [datetime(2024,1,25), datetime(2024,2,28), ...] # 2024-2026
RETAIL_SALES_DATES = [datetime(2024,1,17), datetime(2024,2,15), ...] # 2024-2026
ALL_MACRO_EVENT_DATES = sorted(set(CPI + GDP + RETAIL))
CalendarFeatureCalculator(enable_macro_modulation=True)
# In _hour_weight():
base_weight = ... # existing market hours logic (0.1 - 1.0)
if enable_macro_modulation:
# +20% on CPI/GDP/retail sales days
if ts_date in self._macro_event_dates:
base_weight *= 1.20
# VIX regime scale (cached 4h via OpenBB FRED)
vix_scale = _get_vix_regime_scale() # 0.9x to 1.3x
base_weight *= vix_scale
| VIX Level | Scale | Rationale |
|---|---|---|
| < 15 | 1.0x | Normal volatility |
| 15-25 | 0.95x | Slightly reduce weight (elevated but manageable) |
| 25-35 | 1.1x | Higher volatility = more signal in price moves |
| > 35 | 1.3x | Panic = high weight (big moves, high importance) |
_macro_data_cache dict (not instance-level)_fetch_vix_level() using OpenBB FRED_macro_event_dates pre-built as set in __init__ for O(1) lookup| Attempt | Why it Failed | Lesson Learned |
|---|---|---|
| Adding new feature (earnings_proximity) | Changes obs_dim from 5900 to 6000, breaks all existing models | Modulate existing features instead |
| Instance-level VIX cache | Multiple CalendarFeatureCalculator instances waste API calls | Module-level cache shared across instances |
MACRO_EVENT_BOOST = 1.20 # +20% on macro release days
VIX_REGIME_SCALES = {
'normal': 1.0, # VIX < 15
'elevated': 0.95, # VIX 15-25
'high': 1.1, # VIX 25-35
'panic': 1.3, # VIX > 35
}
VIX_CACHE_TTL = 14400 # 4 hours
| File | Change |
|---|---|
alpaca_trading/features/calendar_features.py | Added date lists, _fetch_vix_level(), _is_macro_event_day(), _get_vix_regime_scale(), modulated _hour_weight() |
tests/test_openbb_integration.py | 6 tests for macro modulation |