Praxis · #372649
Validation · #372649 · Volatility

VIX-ETN vol-premium

The Volatility Edge — A Dual Approach for VIX ETNs Trading, 2025

Carlo Zarattini, Andrew Aziz & Antonio Mele · Swiss Finance Institute
The rule

Harvest the volatility risk premium with short-term VIX-futures ETNs: a dual signal — expected VRP (VIX − 10-day realised vol) and the VIX term-structure slope (VIX vs VIX3M) — sets long-vol / short-vol / cash, sized at VIX/100.

Validated 1.001.03 claimed → measured Sharpe
1.03
Measured (daily)matches the paper's convention
0.87
Honest Sharpemonthly, tail-aware
16.9%
CAGR reproducedequity β 0.11 · α t = 3.3
AssetVolatility
StrategyDual-signal short-vol
UniverseVIX short-term futures ETNs
Period2008–2025
Costs5 bps + ETN fee
InstrumentsVXX VIXY XIV SVXY ^VIX ^VIX3M SPYTraded via short-term VIX-futures ETNs. We reconstruct the underlying S&P 500 VIX Short-Term Futures Index from public CBOE VX settlements (the paper used paid Norgate data) — validated against VIXY and the Feb-2018 +97% reference.

The exact rules

Premium present (eVRP > 0) AND contango (VIX < VIX3M)Short vol — size = VIX/100
Premium gone (eVRP ≤ 0) AND still contangoShort vol — half size (0.5 × VIX/100)
Premium gone (eVRP ≤ 0) AND backwardation (VIX > VIX3M)Long vol — size = VIX/100
Premium present (eVRP > 0) AND backwardationCash — step aside

The backtest, re-run

Inside the model

Position mix
89% short-vol · 7% cash · 4% long-vol
Avg capital deployed
18% (bet size = VIX/100)
Win rate
55% of days · 67% of months
Skew / kurtosis
+1.97 / 58 — fat tails, but the big moves are UP (signals flip long-vol in stress)
Best / worst day
+20% / −13%
Best / worst month
+60% / −8%
Return concentration
made in high-premium years (2008 +78%, 2020 +46%); ~flat when calm (2014 +1%, 2024 +2%)
Recent (decay watch)
2024 +2%, 2025 +6% (partial) — possible VRP compression

The validation ladder

C0ReplicateAll 4 rule sets match — S1 Sharpe 0.49 vs 0.48 … S4 1.03 vs 1.00, CAGR 16.9% vs 16.3%, MDD −30.9% vs −31%.
C1HonestyAdd the ~0.95%/yr ETN expense ratio the index backtest omits → net daily Sharpe 1.02 (barely dented). Autocorrelation-honest monthly Sharpe: 0.87.
C2DeflateOn non-overlapping monthly returns (daily √252 flatters a carry book): survives BHY multiple-testing (t = 3.85) and the Deflated Sharpe — statistically real.
C3CrisisSidestepped Volmageddon (term-structure flattened the book the Friday before). Real max drawdown −30.9% came in COVID-2020 instead; rates-2022 −11%.
C5FrictionsCost sweep 0bp → 1.04, 5bp → 1.02, 10bp → 1.00, 20bp → 0.95, 40bp → 0.86 — robust, not a cost mirage.
C6DecayIS 1.16 → OOS 0.90 → recent 1.05 — minimal decay (but a 2025 paper has no true post-publication out-of-sample).
C7Originalityvs SPY: α 15.5%/yr (t = 3.3), β 0.11, R² 0.02 — genuinely uncorrelated. The 'alpha' is the volatility risk premium: real and compensated, not an anomaly.
ReproducesReal after costsStatistically realCrisis-robustOriginal vs equitiesOut-of-sample proven

How we rebuilt it

Data
S&P 500 VIX Short-Term Futures Index reconstructed from public CBOE VX settlements + Yahoo ^VIX/^VIX3M/SPY (paper used paid Norgate ETN data).
Method
Dual signal (expected VRP + VIX term-structure slope) → long-vol / short-vol / cash, sized at VIX/100; 3:45pm MOC, ±2% band.
Universe
Short-term VIX-futures ETNs (VXX / XIV / SVXY-type), 2008–2025.
Regime stated
None — but the term-structure signal is itself the regime filter (it flattened the book before Volmageddon).
Deviations from the paper
  • Data: paper's Norgate ETN series (paid) → S&P VIX Short-Term Futures Index reconstructed from public CBOE VX settlements; validated vs VIXY (annual within ~2pp) and Volmageddon (+97% exact).
  • Index/NAV (settlement) basis — matches XIV's −96% NAV and Norgate ETN data, not the 3:45pm intraday price.
  • C0 at the paper's 5 bps; C1 adds the ~0.95%/yr ETN expense ratio the index backtest omits.
  • Sharpe reported daily (1.03) to match the paper; the autocorrelation/tail-honest monthly Sharpe is 0.87 — and the dynamic sizing does NOT beat the fixed-size variant risk-adjusted (S3 monthly 0.92 ≥ S4 0.87), it just adds CAGR + tail.

ProvenanceBeta

Engine
v1
Blocks
5 new (vix + vix_futures reconstruction, eVRP, term-structure, banded backtest), 5 reused (metrics, G1/G2, regime, spanning)
Data
cboe_vx_recon/sp_st_futures_ER · Jan 2008 – May 2025 · run_id 5
Source
CBOE VIX futures settlements ↗ · S&P VIX Futures Index methodology ↗
Tests
21/21 passing · reconstruction known-value + strategy-logic + adversarial-code QA (3 findings fixed)
Reproduce
view code ↗
Validation Report · #372649

The Volatility Edge — A Dual Approach for VIX ETNs Trading, 2025

One email opens this reproduction in full, and the rest of the library: ten systematic strategies, re-run and audited. Free.

Descriptive research, not investment advice. No card, no trial clock.
© MMXXVI MarketGraphs · descriptive research, not investment advice