GEX as a Predictive Tool: Why VIX Explains Most of Its Performance

Options traders often rely on dealer-exposure metrics like gamma exposure (GEX) to forecast market volatility and returns, assuming these signals offer unique predictive power. However, a rigorous eight-year backtest of GEX and related metrics against SPY data suggests that most of their apparent value stems from their correlation with the VIX rather than independent forecasting ability.

The Role of Dealer-Exposure Metrics in Options Trading

Dealer-exposure metrics—such as GEX, delta exposure (DEX), vega exposure (VEX), and charm exposure (CHEX)—are frequently marketed as advanced tools for predicting market movements. The premise is that dealers hedge their positions, creating flows that influence volatility, returns, or implied volatility (IV) changes. While the mechanics behind these metrics are valid, their predictive utility remains a subject of debate.

To test this, a backtest was conducted over 1,972 trading days from April 2018 to April 2026, analyzing SPY’s end-of-day (EOD) snapshots. The study evaluated whether these metrics provided incremental predictive information beyond what is already available from the VIX and ATM implied volatility.

Raw GEX Backtest Results: A Strong but Misleading Signal

When SPY days were sorted into quintiles based on GEX, the results appeared compelling. The most negative GEX quintile (Q1) showed a mean next-day realized volatility of 16.97%, while the most positive GEX quintile (Q5) had a mean of just 6.34%. The difference of -10.63 percentage points was statistically significant (t = -13.00, p = 1.0e-33), with a Spearman rank correlation of -0.36.

However, this raw performance may be misleading. High negative GEX often coincides with elevated VIX levels, suggesting that GEX might simply be acting as a proxy for volatility regimes rather than an independent predictor. To isolate GEX’s true predictive power, the study controlled for VIX and ATM IV.

Incremental Predictive Power Fades After Controlling for VIX

After regressing GEX against VIX alone, its predictive power dropped dramatically. The Spearman correlation for GEX’s effect on next-day realized volatility fell from -0.36 to -0.14, and after adding ATM IV to the regression, it further declined to -0.03 (p = 0.18). This suggests that GEX’s apparent edge is largely explained by its relationship with VIX and IV, rather than offering unique forecasting value.

The same pattern emerged for other metrics:

• DEX: Showed no meaningful predictive power for next-day returns after controls (ρ = +0.02, p = 0.40).
• VEX: Initially appeared to predict changes in ATM IV (ρ = -0.16, p = 2.1e-13), but this effect vanished after accounting for VIX and IV (ρ = -0.01, p = 0.77).
• CHEX: Displayed a slight raw correlation with next-day returns (ρ = -0.05, p = 0.03), but this disappeared entirely under VIX control (ρ = -0.01, p = 0.63).

The study’s pre-registered tests confirmed that none of these metrics provided robust incremental predictive information beyond VIX and ATM IV.

Double-Sort Analysis Reveals GEX’s Limitations in High-Volatility Regimes

To further investigate GEX’s behavior, the data was double-sorted by VIX quintiles and GEX quintiles. In low-to-moderate VIX regimes (V1-V3), positive GEX was associated with lower realized volatility, as expected. However, in the highest VIX quintile (V5), the relationship broke down entirely. The mean realized volatility did not decrease monotonically with positive GEX, and in some cases, high-GEX days saw higher volatility than their low-GEX counterparts.

This inconsistency suggests that GEX’s predictive power is strongest in calm-to-moderate market conditions but fails to provide clear signals during extreme volatility, where forecasting edges are most valuable.

Practical Implications for Traders

The findings underscore a critical reality for options traders: dealer-exposure metrics like GEX are not inherently flawed, but their perceived predictive power is largely a reflection of broader volatility regimes captured by the VIX. While GEX can serve as a useful regime descriptor—helping traders identify whether dealers are net long or short gamma—it does not reliably outperform simple VIX-based strategies in forecasting next-day volatility or returns.

For traders seeking genuine predictive edges, combining VIX with other indicators—such as order flow, sentiment, or macroeconomic factors—may prove more fruitful than relying solely on dealer-exposure metrics. The backtest results highlight the importance of rigorous statistical validation before adopting any quantitative trading strategy, especially one marketed as a "silver bullet" for volatility prediction.