VIX DMA Edge: probability exits, Kelly sizing, shorting (qm-a3a)
Rejected qm-a3a ·
Decision: NO-SHIP
No variant cleared the improvement bar versus researcheddefaults (cashonupper + meanortime, edgefreq=0.02, time=5d). Catalog defaults and VixDmaEdge.researcheddefaults() stay unchanged.
Provenance
Equity vs S&P (aligned, rebased to 1.0)
Aligned window, both rebased to 1.0 at start — Aligned window, both rebased to 1.0 at start for fair comparison · S&P dashed, experiment solid. X = Date, Y = Equity (rebased to 1.0).
Signal driver & thresholds (x-aligned)
VIX DMA distribution
Probability exits, Kelly sizing
Stats
| CAGR | 7.45% | Sharpe | 68.90% |
|---|---|---|---|
| Sortino | 100.85% | maxDD | -17.03% |
Disposition
# VIX DMA Edge: probability exits, Kelly sizing, shorting (qm-a3a) ## Decision: **NO-SHIP** No variant cleared the improvement bar versus `researched_defaults` (cash_on_upper + mean_or_time, edge_freq=0.02, time=5d). Catalog defaults and `VixDmaEdge.researched_defaults()` stay unchanged. **Follow-up (qm-a3a.1):** variant parity audit confirms params are not no-ops (weight series differ) but metrics are genuinely similar (small Kelly stakes); monthly SPY corr ≈ 0.99. See `vix-dma-edge-audit-qm-a3a.1.md`. ## Hypothesis 1. **Probability→exit**: Expanding-window P(|dev| shrinks over H days |
Follow-on work
- Follow-up qm-a3a
- Follow-up qm-a3a.1
- Follow-up qm-a3a.1.md
Errors & data gaps
- Live SH from Yahoo via `ensure_ohlcv` (IPO ≈ 2006-06-21). - Research ticker `__SYNTH_SH_INV_SPY__`: compound −1× SPY daily simple returns; `splice_sh_with_synthetic` level-matches at first live SH bar. - Not upserted into tracked `market.sqlite` (budget); built in-panel for sweeps/tests. Follows `__…__` synthetic naming conventions.
Follow-up
gh issue create --title "Follow-up VIX DMA Edge probability exits Kelly sizing shorting qm-a3a" --body "Follow-up to qm-a3a: Decision: NO-SHIP" --label "priority:p2" --label "pipeline:research" --label "agent:ready"