[Research] Re-run All GEX Research with Expanded Data (2008+, Multi-Ticker)

[iAmGiG]

Objective

Re-run all 8 GEX research strategies with expanded historical data (back to 2008) and broader ticker universe once the EOD backfill completes. This enables analysis across multiple market epochs (GFC, COVID, 2022 bear) and validates whether findings generalize beyond SPY.

Blocked by: EOD historical backfill (currently in progress — equity prices from 2008, options chains, regime data for all symbols).

---

Current Data State (as of 2026-02-16)

Table	Current Coverage	After Backfill
equity_prices_daily	61 symbols, 2020–2026	61+ symbols, 2008–2026
options_daily_summary	SPY: 1,505 days (2020–2025); QQQ/IWM: ~224 days	Full regime data for all symbols
options_chains_*	Empty (backfill in progress)	Full chains 2008–2026
intraday_gex_snapshots	Does not exist yet	TBD

Current Symbol Universe (61 tickers)

SPY, QQQ, IWM, XLF, XLE, XLK, XLV, AAPL, MSFT, AMZN, GOOGL, META, NVDA, TSLA, and 47 others in equity_prices_daily.

Key Limitation

Regime data (options_daily_summary.regime) currently only has meaningful coverage for SPY. Multi-symbol regime stratification requires the backfill to populate regime classifications for QQQ, IWM, sector ETFs, and individual names.

---

Research Tasks to Re-run

Batch 1: Core GEX Strategies (originally 2021–2023, SPY-focused)

• [Research] Delta Neutral / Long Volatility Signal from GEX Data #544 Delta Neutral / Long Volatility — Re-run with 2008–2023 data

  • Test whether vol thresholds stabilize with longer calibration windows
  • GFC 2008 + COVID 2020 provide extreme-vol calibration data
  • Multi-symbol: QQQ, IWM (if regime data available)

• [Research] Correlation / Dispersion Trading via GEX #545 Correlation / Dispersion Trading — Re-run with expanded sectors

  • Currently: SPY vs XLF/XLE/XLK/XLV/QQQ/IWM
  • Add: individual mega-caps (AAPL, MSFT, NVDA, etc.) as correlation pairs
  • Test 2008–2023 for GFC dispersion behavior
  • Best candidate for improvement — already showed +0.108 Sharpe improvement

• [Research] Pairs Trading (Mean Reversion) Z-Score Signal #546 Pairs Trading Mean Reversion — Test new pairs

  • SPY/IWM failed cointegration — try sector ETF pairs (XLK/XLF, XLE/XLV)
  • Individual stock pairs from same sector
  • Longer data (2008+) may reveal cointegration that shorter windows miss

Batch 2: Extended Research (originally 2020–2023)

• [Research] Re-validate MACD+RSI in Positive Gamma with Corrected Methodology #531 MACD+RSI Regime Re-validation — Expand symbols and period

  • Current: SPY Sharpe 7.73 bullish vs -4.52 bearish (p<0.0001) — strongly confirmed
  • Re-run for QQQ, IWM, XLK, XLF, and individual names
  • Test 2008–2023 to see if regime effect persists across market epochs

• [Research] Intraday GEX Validation - Practitioner Use Case #530 Intraday GEX Validation — Re-test with more data

  • Current: No monotonic relationship (Spearman p=0.39), U-shaped pattern
  • More data may reveal signal in specific regimes or market conditions
  • Multi-symbol validation

• [Research][Testing] Transaction Cost Sensitivity Analysis #541 Transaction Cost Sensitivity — Re-sweep with longer backtest

  • Current: Pairs breakeven at 5.9 bps; dispersion/delta neutral negative at all costs
  • Longer data = more trades = more robust breakeven estimates
  • Test whether GFC and COVID periods change the cost sensitivity profile

• [Research][Testing] Real Market Fat Tails Validation #542 Fat Tails Validation — Add GFC period

  • Current: kurtosis=10.15, Conservative optimal full-period
  • GFC 2008–2009 is the ultimate fat-tails test — should produce extreme kurtosis
  • Compare crash behavior across GFC vs COVID vs 2022 bear

• [Research] VaR vs CVaR Risk Metrics for RiskAgent #543 VaR/CVaR Risk Metrics — Regime-stratify with more data

  • Current: ES/VaR=1.69 full sample; bearish gamma VaR 4.3x higher than bullish
  • More regime data for more symbols enables cross-asset risk comparison
  • GFC tail behavior should push ES/VaR even higher

---

New Analysis Opportunities (with 2008+ data)

Market Epoch Comparison

Epoch	Period	Character
GFC	2008-09 to 2009-03	Systemic financial crisis
Post-GFC Recovery	2009-03 to 2012-12	Low-vol grind higher
Pre-COVID Bull	2013-01 to 2020-02	Extended bull market
COVID Crash	2020-02 to 2020-03	Fastest bear market in history
COVID Recovery	2020-03 to 2021-12	Stimulus-driven rally
2022 Bear	2022-01 to 2022-10	Rate-driven bear
2023-2025 Bull	2023-01 to 2025-12	AI-driven bull

Key Questions

1. Does the GEX regime overlay add more value during crisis periods vs calm markets?
2. Are the MACD+RSI regime differences consistent across all epochs?
3. Do fat-tail risk metrics (ES/VaR) show structural changes over time?
4. Which strategies improve most with longer calibration windows?
5. Do any strategies that failed on 2020–2023 work on 2008–2020?

---

Backfill Progress Checklist

Before re-running, verify:

• equity_prices_daily has data back to 2008 for core symbols
• options_daily_summary has regime data for QQQ, IWM, sector ETFs
• options_chains_* partitions populated for 2008–2020
• Run SELECT symbol, MIN(trading_date), MAX(trading_date), COUNT(*) FROM equity_prices_daily GROUP BY symbol ORDER BY symbol to verify coverage

---

Implementation Notes

• All research runners are in scripts/research/
• Signal generators in src/backtesting/signals/
• Backtester: src/backtesting/research_backtester.py (PostgreSQL-backed)
• GEX overlay: src/backtesting/gex_overlay.py
• Risk metrics: src/backtesting/risk_metrics.py
• Existing test suite: 121 unit + 33 integration = 154 tests

Runner Modifications Needed

• Parameterize start/end dates (currently hardcoded to 2020–2023)
• Accept ticker list from CLI args or config
• Add epoch-splitting logic for cross-epoch comparison
• Output per-epoch YAML results alongside full-period results

---

Related Issues

• [Research] GEX Regime Overlay — Consolidated Findings (#544, #545, #546) #547 — Consolidated GEX research findings (current results)
• [Research] Delta Neutral / Long Volatility Signal from GEX Data #544 — Delta Neutral / Long Volatility (v1 complete)
• [Research] Correlation / Dispersion Trading via GEX #545 — Correlation / Dispersion Trading (v1 complete)
• [Research] Pairs Trading (Mean Reversion) Z-Score Signal #546 — Pairs Trading Mean Reversion (v1 complete)
• [Research] Intraday GEX Validation - Practitioner Use Case #530 — Intraday GEX Validation (v1 complete)
• [Research] Re-validate MACD+RSI in Positive Gamma with Corrected Methodology #531 — MACD+RSI Regime Analysis (v1 complete)
• [Research][Testing] Transaction Cost Sensitivity Analysis #541 — Transaction Cost Sensitivity (v1 complete)
• [Research][Testing] Real Market Fat Tails Validation #542 — Fat Tails Validation (v1 complete)
• [Research] VaR vs CVaR Risk Metrics for RiskAgent #543 — VaR/CVaR Risk Metrics (v1 complete)

[iAmGiG]

Status (March 2026): Low priority. Depends on expanded historical data (2008+). Also absorbs #558 (Gamma Distribution Shape Analysis) concept.