**Source:** golden-firmament:phase3-results
**Time:** 2026-06-13T13:32:28
---
## GOLDEN FIRMAMENT β Phase 3 Complete (Optimization)
**From:** TZEDEK
**Status:** optimized strategies identified
**Timestamp:** 2026-06-13T13:35+00:00
### Winner 1: Solar Calendar (SPY) β Invert Sell in May
- Sharpe: 1.25 | WR: 69.2% | PF: 2.57 | Tot: +194.7% (26yr)
- Fixed calendar dates, no daily input needed
- Excluding Ides of March pushes Sharpe to 1.38 with PF 3.17
### Winner 2: Lunar Phase Swing (GLD) β +2% Stop Loss
- Sharpe: 1.18 | WR: 51.4% | PF: 2.10 | Tot: +97.3%
- Gold responds strongly to lunar cycles (better than SPY)
- Stop loss at 2% dramatically improves risk metrics
### Runner-up: Lunar Phase Swing (SPY) β Baseline
- Sharpe: 0.68 | WR: 64.4% | PF: 1.59 | Tot: +84.8%
- Decent standalone, outperformed by GLD variant
### Key Findings
- RSI filter on Lunar: REDUCED performance (removed valid entries)
- Stop loss on Lunar: 2% SL on GLD = best risk-adjusted variant
- Sell-in-May inversion: Best single optimization on Solar Calendar
- USO: Not suitable for either strategy (too volatile/correlated with oil shocks)
### Files
- /root/quant/golden-firmament/phase3_optimize.py β full Phase 3 script
- All results computed across SPY, QQQ, GLD, USO (5 variants Γ 4 tickers = 20 runs)
### Next for Phase 4 (Final)
1. Synthesize report
2. TZEDEK delivers to Marko via Telegram