Date: 2026-03-21 Status: Implemented
MS8h strategy backtested per-coin in isolation shows +130% to +427% over 14 months. But the portfolio-level question — which coins, when to switch, what size, which trail stop — was never comprehensively tested.
Single simulation engine testing every combination across 8 dimensions:
| # | Dimension | Values | Count |
|---|---|---|---|
| 1 | Selection method | regime_avg, blend, top_pnl, top_wr, random, fixed | 6 |
| 2 | Portfolio size | 4, 6, 8, 10, 12, 15, 20 | 7 |
| 3 | Swap strategy | never, swap_1, swap_2, swap_all, full_rotation | 5 |
| 4 | Swap friction | none, realistic | 2 |
| 5 | Direction | both, long_only | 2 |
| 6 | Capital allocation | equal, score_proportional | 2 |
| 7 | Trail stop | 0.003, 0.005, 0.010 | 3 |
| 8 | Rebalance trigger | regime_change, weekly, monthly | 3 |
Total: ~15,120 deterministic configs + 12,600 random repetitions.
experiments/ms8h_live/portfolio_sim.py (1,678 lines) — engine + config generationexperiments/ms8h_live/portfolio_sim_report.py — 4-layer markdown + 7 Chart.js charts