S032 forward-paper track — buy-write DIA Δ30 + QQQ Δ30 (real chains)
Setup: S032 (buy-write-rsi2-etf-v1) · Stage: 3 (forward-paper) · Started: 2026-07-02
Status: PRE-REGISTERED and LIVE-COLLECTING. Frozen on first booked trade. Public. Not live/promoted — the forward record must clear the inherited gate before any live claim.
Why forward, and why only two cells. The modeled Stage-1 screen passed on flat-ATM implied volatility (IV), which overstates the real (skewed) premium — most for the lower-delta cells. Only real chains settle it. We forward-track the two least-distorted, highest-Sortino-lift cells — DIA Δ30 and QQQ Δ30 — a blue-chip vs high-vol contrast on one delta. Sample is thin (one underlying per cell, ~8–10 signals/yr), so the read is directional and accumulates slowly, in the open.
Frozen protocol (the base trade — a changed parameter is a new track)
- Cells: DIA Δ30, QQQ Δ30. Single-cycle, hold-to-expiry (as S032 — not the S033 rolling variant).
- Signal: on the last completed EOD session,
RSI-2(close) < 10 AND close > 200-DMA(the S027 trigger). Spacing ≥ 5 trading days between entries per cell; cap 4 open per cell (a new signal at 4/4 is skipped, not queued). - Entry execution: during US market hours on the next session (≈ the "open(T+1)" of the backtest). Buy 1 unit of the ETF at the live mid; write the ~30-DTE monthly call whose live delta is nearest 0.30. The call is sold at the bid (worse-than-mid, conservative).
- Exit: at the call's expiry — record the ETF spot near the close; the call settles at intrinsic
max(spot − strike, 0). If the ETF ≥ strike the unit is called away at the strike (capped); else the premium is kept. - P&L (per trade):
[min(exit_spot, strike) − entry_spot + premium] / entry_spot − 30 bps(ETF round-trip). The matched naked long ((exit − entry)/entry − 30 bps) is booked on every trade as the baseline. - Friction: ETF 30 bps; the option is sold at the bid (no mid credit). Recorded, never flattered.
- Capture: real strike, bid/ask/mid, real delta and IV, open interest, expiry — so realised skew and fills are what the market actually gave, replacing the flat-IV assumption.
Gate (inherited from S032 §9 — frozen, risk-adjusted, tail-aware)
Applied to the realised forward record, per cell and pooled:
- Primary: per-trade Sortino(buy-write) > Sortino(matched naked long) on real premium. The whole forward question is whether the real (skewed, bid-filled) premium still clears the naked long the way the flat-IV model claimed — or whether the skew haircut erases it.
- Tail (co-gate): worst trade and conditional value-at-risk (CVaR) reported; the buy-write must not draw down worse than the matched naked long.
- Raw return reported for context, never the deciding bar (the buy-write surrenders upside by design).
Promotion horizon (FROZEN 2026-07-03): no promotion decision before ≥ 30 completed trades (pooled across the two cells) OR ≥ 12 months of collection — whichever comes first. Until that mark the read is explicitly directional, and nothing is promoted or called live. At the mark the inherited gate (per-trade Sortino > matched naked long, tail-aware) is applied to the realised record and the decision (promote to live consideration / cut / extend) is taken in the open. Collection began 2026-07-02, so the time bound falls 2027-07-02 absent 30 trades first.
Positioning (public, honest)
Low-alpha by design, risk-reducing not risk-free. A variance-risk-premium harvest that shaves drawdown and adds a small risk-adjusted edge over owning the ETF, paid for with capped upside. It keeps the ETF's full downside below the thin premium cushion — income, not tail protection. Forward-paper, transparent, pre-registered; live only if the real-chain record clears the inherited gate.
Collector: the engine (entries during market hours, exits at expiry) → table buy_write_forward_events. Public page: /lab/buy-write-rsi2-etf-v1/forward. Frozen on first booked trade; a changed parameter spawns a new forward track. English per repo convention. Modeled screen: the internal record. Gate inherited from the S032 spec §9.