← AI features

Factor inventory — model card

v2.7 · Live The 30+ factor library powering the scoring engine. Tier 1-4 hierarchy: macro → sector → instrument → sentiment. Per Bible §6.1 inventory must be explicit.

Tier 1 — Macro / cross-asset (always-on; weight ≥ 0.30)

IDNameSourceNotes
F01_macro_moodMacro mood (composite of DXY/yields/VIX)FRED + YahooDrives regime layer
F02_dxyDollar index trend + levelFRED DTWEXBGSRisk-on inverse correlator
F03_yield_curve10Y-2Y spreadFRED T10Y2YRecession proxy
F04_creditHY credit spread (OAS)FRED BAMLH0A0HYM2Risk appetite
F09_vixVIX level + slopeYahoo ^VIXVol regime classifier

Tier 2 — Sector / industry (weight ≥ 0.20)

IDNameSourceNotes
F11_sector_rotationSector momentum cross-sectionYahooTop-3 / bottom-3
F12_breadth% above 200d SMA in sectorYahooInternal participation
F13_intermarketSPY/TLT/GLD/USD correlationYahooRisk-on confirmation
F15_vol_regimeSector-level vol percentileYahooPosition sizing input
F18_relative_strengthSector RS vs SPYYahooRotation signal

Tier 3 — Instrument fundamentals (weight ≥ 0.15)

IDNameSourceNotes
F22_earningsEarnings revision + surpriseFinnhub / SECQuality factor
F23_liquiditySpread + book depthYahoo + CoinbaseBelow 25 → no-action
F24_short_interestSI as % of floatFinnhubSqueeze setup signal
F25_squeezeSqueeze composite (SI + days-to-cover + vol)ComputedLead indicator
F26_insiderInsider net buys (last 90d)SEC EDGARSignal-rich for < $5B mcap
F28_qualityROIC + FCF + balance sheetFMPRisk-off rewarded

Tier 4 — Sentiment / technical (weight ≥ 0.05)

IDNameSourceNotes
F05_mom3m3-month momentumBarsRisk-on rewarded
F06_mom1m1-month momentumBarsFaster-decaying
F07_mean_reversionZ-score from 50d meanBarsCounter-trend
F08_news_velocityNews story rate × sentimentNews aggregatorEvent-window detector
F10_atr_vol14-day ATR % of priceBarsCompression detector
F31_fundingCrypto perp funding rateBinance Futures (v77)Crypto-only
F32_oi_changeOpen-interest deltaBinance Futures (v77)Crypto-only
F33_smart_moneyWhale + tier-1 fund flowsSEC 13F + on-chainQuarterly lag

Layer-conditional weighting

Per engine/v76-layer-conditional.js, weights vary by regime:

When to add a factor (Bible §6.6)

Proactively. When auditing API responses, watching the news cycle, or reading research:

  1. Add to FactorRegistry
  2. Backfill historical data
  3. Re-weight if necessary (sum to 1.0 ± 0.05)
  4. Add unit test (factor produces values in expected range)
  5. Document here + in HANDOVER.md
  6. Bump engine VERSION (Bible §15.7)

Source

engine/factors.js, engine/factors-extended.js, engine/factors-extended-2.js, engine/factors-extended-3.js, engine/v33-improvements.js through engine/v52-cockpit-views.js.

Reviewed: 2026-04-27 · Next: 2026-07-27 · Per Bible §15.61 + §6.1