BtcDeribitWeeklyShortStrangleVrpIvRankGated

BTC Deribit Short-Strangle Variance Risk Premium Harvest (Weekly, IV-Rich Gated, Delta-Balanced Market-Neutral)

First options strategy in the factory portfolio (options venue currently 0% vs >=15% quota). A market-neutral short-volatility strategy on BTC Deribit European options that harvests the persistent crypto variance risk premium: BTC implied volatility trades structurally richer than subsequently realized volatility because perpetual demand for convex upside calls and crash-protection puts keeps option prices elevated. Each week the strategy sells a delta-balanced ~16-delta OTM strangle (one OTM call + one OTM put on the same near-dated weekly expiry) ONLY when the vol risk premium is demonstrably positive (ATM IV elevated vs trailing realized vol AND IV in the upper portion of its own trailing distribution), so it stands down in cheap-vol regimes rather than blindly selling premium. Deribit options are European-style (no early-assignment risk) and BTC is the most liquid crypto options market with the deepest IV/greeks history in the catalog. Weekly rolling gives ~40-52 trade cycles per year, populating walk-forward and holdout windows (directly addressing the event-sparsity failures seen in rare-event spot strategies). This is NOT a directional bet and NOT dependent on any thin/short-history supplementary stream (liquidations/OI/L-S ratios) or on the funding stream that ends 2024-12 — it reads option IV/greeks and underlying OHLCV only.

Implements the hypothesis exactly: the first options strategy in the portfolio — a single-venue (DERIBIT), options-only, delta-balanced weekly short strangle gated on IV>RV AND IV-rank. Edge is the structural crypto Variance Risk Premium (buyers overpay for convex calls + crash puts), monetized two-sided and market-neutral — not directional, not a funding/basis carry, and using NO thin/short-history supplementary stream: it reads only deribit_dvol (continuous 2021->2026 BTC implied-vol index; deribit_book_summary mark_iv as fallback) and underlying OHLCV. RV is annualized log-return vol over the trailing 7 days (168 1H bars); IV-rank is the percentile of current IV within a ~60-day window. The dual gate (IV >= RV*vrp_margin AND IV-rank >= threshold) only sells vol when it is rich in both absolute and relative terms, so it stands down in cheap-vol regimes. Symmetric +-7% offsets place the shorts near ~16-delta, netting delta ~0 without any hedge leg (strictly single-venue). Weekly turnover (~40-52 cycles/yr) densely populates walk-forward + holdout windows (addressing event-sparsity) and keeps the ~0.06%RT x2 Deribit fee negligible vs 1-3% weekly premium. Routed to DERIBIT (deepest BTC option liquidity / IV history); leverage=1 (options venue cap) with BTC-contract risk_frac sizing, so no unused-leverage gate triggers. Fills the 0%-vs->=15% options portfolio quota.

Redundant duplicate. This backtest is byte-for-byte identical to the already-reviewed-and-advancing sibling BtcDeribitWeeklyShortStrangleVrpGated (8ab5ff3d): same total_return (2.1624), Sharpe (4.7136), 98 trades, identical daily_returns and options_summary (net_theta 868.68, net_vega -262.37, net_delta -0.182). The only code change is an added IV-rank gate, and it is demonstrably NON-BINDING — it yields exactly the same entries, exits, and every metric as the ungated sibling, so it contributes zero differentiation or new edge. The sibling was already sent to optimization in the prior turn; optimizing this twin would burn ~2 hours re-evaluating the identical VRP short-strangle while testing a gate that does nothing in this sample. The right move is to concentrate the VRP short-strangle evaluation on the single sibling already in the pipeline (where the deflated-Sharpe/holdout/tail gates will adjudicate the ~1-year, regime-limited sample and the ~65% short-vol drawdown masked by the headline Sharpe 4.71 / 90% win-rate); if the IV-rank filter is worth testing, it belongs as a parameter inside that run, not as a second identical pipeline. FAILURE PATTERN: two near-identical short-strangle VRP strategies submitted in the same session whose only difference (an extra IV-rank gate) is non-binding and produces a byte-identical backtest — abandon the duplicate rather than spend optimization compute twice on the same edge.

Single-venue, options-only, market-neutral VRP harvest on BTC Deribit weekly European options. Each week, when BOTH (a) near-term implied vol (DVOL index) exceeds trailing 7-day realized vol by a margin AND (b) IV-rank is in the upper part of its trailing distribution, it sells a delta-balanced ~16-delta OTM strangle on the nearest weekly expiry (short call ~+7% strike + short put ~-7% strike, symmetric so net delta ~0 with no perp hedge). Both premiums are collected and held to cash-settled expiry, closed 24h before to avoid settlement slippage, capturing theta decay. In cheap-vol regimes the gate stays shut and it stands down. Built on RollingOptionsStrategy: the Deribit perp (primary) drives the weekly roll clock and supplies 1H history for RV; the option legs are injected by the rolling-options chain expander at backtest time. The VRP+IV-rank gate and a once-per-cycle entry window are layered into _enter_leg.

Real mechanism trades as designed (98 short option legs, net_theta +869, short vega, avg_dte 6.9) — a genuine VRP short-strangle harvest, not a degenerate zero-trade options roll

Single-venue / single-account, options-only, fills the under-represented Deribit bucket

Coherent short-vol return signature (negative skew -2.53, avg_loss >> avg_win) with losses properly marked

Byte-for-byte IDENTICAL backtest to the already-reviewed sibling BtcDeribitWeeklyShortStrangleVrpGated (8ab5ff3d) — same return 2.162, Sharpe 4.714, 98 trades, same daily_returns and options_summary

The sole differentiator (IV-rank gate) is NON-BINDING: it produced exactly the same trades/metrics as the ungated sibling, so it adds no edge, no differentiation, and no new information

That sibling was already sent to optimization last turn; re-optimizing this duplicate would waste ~2 hours of compute on the same strategy

Same unresolved concerns as the sibling: only ~1 year / 52 weekly expiries of resolved chain (regime-limited), and Sharpe 4.71 / 90% win-rate masks a real ~65% drawdown (−62.7% week 2026-02-05, −39.3% week 2026-06-04) — short-vol tail illusion

BtcDeribitWeeklyShortStrangleVrpIvRankGated sold weekly delta-balanced OTM strangles on BTC Deribit to harvest the variance risk premium, gated on both IV-vs-RV and IV-rank, as the factory's first options strategy filling the empty Deribit bucket. The mechanism traded as designed (98 real short legs, correct short-vol greeks, headline Sharpe 4.71), but its backtest was identical to an already-advancing sibling because the IV-rank gate was non-binding — it changed nothing in this sample. The analyst ruled it a redundant duplicate, directing the VRP evaluation to the single sibling already in optimization (where the regime-limited ~1-year sample and the masked ~65% drawdown would be adjudicated) rather than spending compute twice. It ended after one iteration as abandoned, never advancing to optimization or risk review.

A new gate that produces a byte-identical backtest to an existing sibling adds zero differentiation and should be a parameter inside that sibling's run, not a second pipeline: concentrate the VRP short-strangle evaluation on the one instance already advancing (where DSR/holdout/tail gates can adjudicate the ~1-year regime-limited sample and the ~65% short-vol drawdown that the headline Sharpe 4.71 / 90% win rate conceals).

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as a redundant duplicate — the only differentiator from the already-reviewed, already-optimizing sibling was an added IV-rank gate that proved NON-BINDING (identical trades and every metric), so it adds no edge and re-optimizing it would burn ~2 hours re-evaluating the same strategy — so optimization and all later stages were never reached.

A market-neutral, options-only short-volatility VRP harvest on BTC Deribit weekly European options — each week selling a delta-balanced ~16-delta OTM strangle (one ~+7% call + one ~-7% put on the nearest weekly expiry) and holding to near-expiry, but only when the variance risk premium gate was open (DVOL implied vol ≥ trailing realized vol × margin AND IV-rank ≥ 50) so it stands down in cheap-vol regimes — the factory's first options strategy, reading option IV/greeks and underlying OHLCV only.

Its backtest was byte-for-byte identical to the already-advancing sibling BtcDeribitWeeklyShortStrangleVrpGated: 98 short option legs (50 calls / 48 puts, net theta +869, net vega -262, delta -0.18, avg 6.9 DTE), total return 2.16, Sharpe 4.71, 90% win rate, profit factor 2.35, negative skew -2.53 with avg_loss >> avg_win — over only ~1 year / 52 weekly expiries of resolved chain, with the headline masking a real ~65% short-vol drawdown (-62.7% week 2026-02-05, -39.3% week 2026-06-04).