The MA drives returns. The VIX protects against tail risk. We test whether adding VIX-based exits to our momentum strategy can reduce drawdowns without sacrificing returns. The answer: a specific VIX filter cuts max drawdown by 30% at a cost of less than 0.5% CAGR. Applied to our momentum strategy on US large-cap equities. 254 months, 2005–2026.
The VIX does not improve the Sharpe ratio. No VIX-based filter, in any configuration tested, produces a higher Sharpe than the MA200 alone. However, one specific combination — adding a “VIX > 25 and rising” exit on top of MA200 — reduces maximum drawdown from -31.1% to -22.3% at a cost of only 0.4pp CAGR. This is the only actionable finding.
Can the VIX replace MA200 entirely? We go to cash whenever the end-of-month VIX closes above a threshold, ignoring the S&P MA200.
| Filter | CAGR | Alpha | Sharpe | Sortino | Max DD | Vol | Cash Mo | $100K → |
|---|---|---|---|---|---|---|---|---|
| MA200 (base) | +17.6% | +9.1% | 0.81 « | 1.19 | -31.1% | 18.0% | 53/254 | $3,092,210 |
| No filter | +18.7% | +10.2% | 0.70 | 1.01 | -55.3% | 22.4% | 0/254 | $3,771,980 |
| VIX <= 20 | +12.4% | +3.8% | 0.58 | 0.75 | -26.5% | 16.2% | 81/254 | $1,177,432 |
| VIX <= 25 | +15.6% | +7.0% | 0.64 | 0.83 | -38.8% | 19.5% | 45/254 | $2,145,765 |
| VIX <= 30 | +16.3% | +7.7% | 0.63 | 0.84 | -40.8% | 20.9% | 22/254 | $2,422,618 |
| VIX <= 35 | +19.4% | +10.9% | 0.76 | 1.05 | -38.0% | 21.6% | 12/254 | $4,293,191 |
| VIX <= 40 | +19.0% | +10.5% | 0.73 | 1.03 | -48.7% | 21.9% | 8/254 | $3,993,136 |
VIX alone is strictly worse than MA200. Best VIX-only Sharpe is 0.76 (VIX ≤ 35) vs 0.81 for MA200. Lower thresholds (VIX ≤ 20) produce excessive cash months (81 vs 53), killing CAGR. Higher thresholds (VIX ≤ 40) leave you exposed to −49% drawdowns. The VIX oscillates too quickly for a monthly rebalance schedule — it triggers false alarms on temporary spikes and misses slow-building downtrends that MA200 catches.
Instead of an absolute level, we test whether the VIX’s direction matters. Invest only when VIX is below its own N-day moving average (fear declining). No S&P MA200.
| Filter | CAGR | Alpha | Sharpe | Sortino | Max DD | Vol | Cash Mo | $100K → |
|---|---|---|---|---|---|---|---|---|
| MA200 (base) | +17.6% | +9.1% | 0.81 « | 1.19 | -31.1% | 18.0% | 53/254 | $3,092,210 |
| VIX < VIX_MA10 | +8.4% | -0.2% | 0.35 | 0.35 | -35.4% | 15.4% | 138/254 | $546,342 |
| VIX < VIX_MA20 | +7.2% | -1.4% | 0.27 | 0.29 | -39.2% | 15.7% | 123/254 | $435,029 |
| VIX < VIX_MA50 | +8.2% | -0.3% | 0.33 | 0.33 | -35.9% | 15.8% | 111/254 | $532,075 |
| VIX < VIX_MA100 | +7.5% | -1.1% | 0.26 | 0.28 | -42.5% | 17.0% | 94/254 | $459,931 |
VIX trend signals are ineffective. Sharpe ranges from 0.26 to 0.49 — all far below the MA200 baseline. The VIX mean-reverts too quickly: by the time it crosses below its MA, the risk event is often already over. Contrarian signals (invest when VIX is rising) fare no better.
Invest only when the VIX’s 1-year rolling percentile rank is below a threshold. This normalises for different VIX regimes across decades.
| Filter | CAGR | Alpha | Sharpe | Sortino | Max DD | Vol | Cash Mo | $100K → |
|---|---|---|---|---|---|---|---|---|
| MA200 (base) | +17.6% | +9.1% | 0.81 « | 1.19 | -31.1% | 18.0% | 53/254 | $3,092,210 |
| VIX pctl <= 50% | +7.8% | -0.8% | 0.32 | 0.32 | -25.9% | 14.9% | 119/254 | $485,821 |
| VIX pctl <= 70% | +9.1% | +0.6% | 0.37 | 0.42 | -37.2% | 16.8% | 79/254 | $637,627 |
| VIX pctl <= 80% | +10.9% | +2.4% | 0.41 | 0.50 | -47.1% | 19.3% | 49/254 | $900,703 |
| VIX pctl <= 90% | +13.5% | +4.9% | 0.52 | 0.66 | -47.7% | 20.1% | 35/254 | $1,453,421 |
Percentile rank is no better. Even at the 90th percentile threshold (only exit when VIX is in its top 10% over the trailing year), Sharpe is just 0.52. The approach spends too much time in cash during normal elevated-VIX periods that precede strong rallies.
The most promising approach: keep MA200 as the primary filter, and add VIX as a supplementary exit signal. Go to cash if MA200 triggers OR if the VIX condition triggers.
| Filter | CAGR | Alpha | Sharpe | Sortino | Max DD | Vol | Cash Mo | $100K → |
|---|---|---|---|---|---|---|---|---|
| MA200 only (base) | +17.6% | +9.1% | 0.81 « | 1.19 | -31.1% | 18.0% | 53/254 | $3,092,210 |
| MA200 AND VIX <= 22 | +11.4% | +2.8% | 0.53 | 0.69 | -28.7% | 15.9% | 80/254 | $976,881 |
| MA200 AND VIX <= 25 | +15.5% | +7.0% | 0.73 | 1.03 | -22.3% | 17.2% | 67/254 | $2,121,446 |
| MA200 AND VIX <= 28 | +15.7% | +7.1% | 0.72 | 1.03 | -31.1% | 17.7% | 57/254 | $2,185,557 |
| MA200 AND VIX <= 30 | +15.7% | +7.1% | 0.72 | 1.03 | -31.1% | 17.7% | 57/254 | $2,185,557 |
| MA200 AND VIX <= 35 | +17.0% | +8.4% | 0.78 | 1.14 | -31.1% | 17.8% | 54/254 | $2,752,321 |
| MA200, skip VIX>22 & rising | +13.7% | +5.1% | 0.66 | 0.88 | -28.7% | 16.2% | 73/254 | $1,504,736 |
| MA200, skip VIX>25 & rising | +17.2% | +8.6% | 0.81 « | 1.17 | -22.3% | 17.4% | 62/254 | $2,858,224 |
| MA200, skip VIX>30 & rising | +16.2% | +7.6% | 0.74 | 1.07 | -31.1% | 17.7% | 56/254 | $2,386,699 |
One combination matches the base Sharpe while cutting drawdown by 9 percentage points.
MA200, skip VIX>25 & rising achieves Sharpe 0.81 (same as base), CAGR +17.2% (vs +17.6%), and max drawdown of −22.3% (vs −31.1%). It requires 62 months in cash vs 53 for the base — 9 extra months over 21 years.
The logic: exit when the VIX is above 25 and rising (above its own 50-day average). This catches the specific scenario of growing fear while ignoring isolated spikes that resolve quickly. The MA200 handles slow downtrends; the VIX condition adds early warning for volatility-driven selloffs.
Do not replace MA200 with VIX. The MA200 remains the superior regime filter for monthly momentum strategies. The VIX oscillates too quickly, produces too many false signals at a monthly frequency, and misses slow-building downtrends.
Optional addition for drawdown reduction: Adding a “skip when VIX > 25 and rising above its 50-day MA” rule on top of MA200 reduces max drawdown from -31.1% to -22.3% at negligible cost to CAGR (0.4pp) and no cost to Sharpe. This is the only VIX configuration worth implementing.