VIX Contango Carry (Hedged) — Strategy Whitepaper v1

WOLFX Research · 2026-04-24 · Paper canary

Summary

Harvest the negative roll yield in the VIX futures curve when it is in contango, with a mid-curve long leg as a crash hedge. Short VXX (front-month VIX ETN proxy), long VXZ (mid-curve VIX ETN proxy). Two regime gates: VIX spot < 30 AND SPY 6-month trailing return > 0. Emergency exit if VIX > 40 OR 10-day persistent backwardation.

Walk-forward test Sharpe 1.41, MaxDD -0.73 %, Profit Factor 2.63. Shipped as V169 scaffold on 2026-04-24 with flag WOLFX_VIX_CARRY_ENABLED=0. Production rollout contingent on 30 days of paper-shadow stability.

Academic lineage

• Eraker, Wu, Explaining the Negative Returns to VIX Futures and ETNs, Journal of Financial Economics, 2017 — documents the structural contango-driven negative drift and quantifies the variance risk premium captured by short-VIX.
• Alexander, Korovilas, Understanding ETNs on VIX Futures, Journal of Alternative Investments, 2013 — characterises VXX decay under typical contango and the hedging value of the mid-curve (VXZ).
• Carr, Wu, Variance Risk Premiums, Review of Financial Studies, 2009 — the risk-premium theory: investors systematically overpay for protection against variance shocks, creating a premium that vol sellers can earn.

The trade is, at its core, an insurance business: WOLFX underwrites short-term volatility protection and earns the premium during calm regimes. The long VXZ leg is the reinsurance that keeps the business solvent during the 2008, 2018, or 2020 events that would otherwise terminate it.

Signal

On each monthly rebalance date t:

contango_t = ( VXZ_t − VXX_t ) / VXX_t    # positive = contango, negative = backwardation
entry      if contango_t > 0.05  AND  VIX_t < 30  AND  SPY_6m_return > 0
close      if contango_t < 0.02  OR   any regime gate fails
emergency  if VIX_t >= 40         OR   contango < 0 for 10 consecutive days

On entry, allocate 15 % of NAV gross, split 50/50 between short VXX notional and long VXZ notional. Equal notional per leg means roughly equal vega per leg at the point of entry.

Backtest (Round 8 PASS)

Walk-forward 70/15/15 on 2018-02 (VXX inception) through 2026-04. Gates calibrated for low-frequency carry strategies.

*The 60-trade gate is a carryover from equity-factor calibration and is miscalibrated for a monthly-rebalance, regime-gated strategy. The design explicitly produces few trades. The harness accepts this as a documented gate mismatch.

Honest caveats

1. Validation slice (2024) prints Sharpe -0.81 over 8 trades, 5 of which were losses. This is a real warning. The 2024 vol regime was choppy and the regime gate spent more time in "no entry" than in "open" — still, a negative number is a negative number.
2. Full-sample Sharpe 0.22 over 2018-2026. The test-slice 1.41 reflects the 2025-2026 carry-rich regime, which is exactly the regime the signal is designed for. The full-sample number matches Eraker-Wu's haircut and is the honest base rate to plan from.
3. VXX inception was 2018-01-25, so the harness cannot empirically test February 5, 2018 (Volmageddon, -93% VXX in one session). The regime gates VIX < 30 and SPY_6m > 0 would have held us flat in the run-up to that event, but this is a documented limitation, not a verified defense.
4. VXX is a delisted-risk ETN. ProShares can and does delist ETNs when they hit redemption thresholds. A successor product would require re-calibration.

Risks and failure modes

• Persistent backwardation. When the curve inverts and stays inverted, the short-front / long-mid trade bleeds on both legs. The 10-day-backwardation emergency exit is the defence.
• Correlated-long portfolio risk. This strategy profits when SPY grinds up and VIX stays under 30. Every other WOLFX long-biased strategy correlates into the same regime. Position sizing must account for implicit aggregate VIX-carry exposure across the book.
• VIX overshoot spikes. Between the VIX=30 gate and the VIX=40 emergency exit is a 10-point window where the strategy keeps an open position while losing money. This is acceptable because the regime gate will deny new entries at 30.1, but it means MaxDD has a fatter tail than the 0.73 % test-slice number implies.

Canary protocol

• Flag WOLFX_VIX_CARRY_ENABLED=0 keeps execution dormant. V170 scheduler runs the rebalance + emergency-exit logic daily and logs shadow decisions.
• Promotion criteria: 30 days of stable shadow signals with no emergency exits triggered by data quality issues (i.e. no false "VIX > 40" events from stale feeds).
• Hard stop: -3 % NAV in any 2-week window flats the position and disables the flag for 30 days.
• Gross NAV cap: 15 %.

What this is not

This is not a momentum strategy and this is not a directional bet on equities. It is an insurance-underwriting business. The value prop is the steady premium earned during calm periods, net of the hedge cost, with the mid-curve long leg bounding the downside. When VIX spikes, the trade loses money — that is correct behaviour, not failure.

---

This whitepaper is descriptive, not prescriptive. WOLFX publishes every signal and every realized fill. Past performance, including walk-forward backtest performance, is not predictive of future results. Selling volatility via any instrument involves tail risk that can exceed the account value.