How many strategies does the live Monte Carlo tracker follow?
7 configured MC strategies (2026-07-07).
Monte Carlo Protocol in 15 Verifiable Facts
Embargo, temporal splits, Sharpe_p5, bootstrap, and validated cells — live VPS figures.
⚠️ Paper trading only. Strategy Arena is a public research lab. Not financial advice. Simulated past performance does not guarantee future results.
How many strategy×asset cells are MC-validated?
36 validated cells in live_mc_results_snapshot.json.
What temporal embargo between MC windows?
48-bar embargo between OOS windows (smart_money_evolved_v2_params.json, May 2026).
What temporal holdout split discipline?
Phase 1: top candidates on 70% train; phase 2: 30 MC OOS splits between 20% and 70% of bars.
Sharpe_p5 threshold for robust validation?
Batch5 filter: Sharpe_p5 > 0.5; 8 cells exceed Sharpe_p5 > 1.0 (best: SNX 1.526).
How many Monte Carlo splits per asset?
30 MC splits by default, minimum 10 valid splits (generic_strategy_optimizer.py).
Minimum simulated trades threshold?
n_trades_mean ≥ 20 trades per validated cell (batch5_mc_optimizer.py).
How many live vs MC aligned cells?
2 aligned cell, 21 major drift (2026-07-07).
How many cells with insufficient live data?
9 cells (< 5 closed live trades).
Flagship MC-validated example?
Smart Money Evolved: 15 validated assets, best Sharpe_p5 1.526 on SNX.
Bootstrap resampling method?
Random OOS anchors + GPU replay of top-train combos on each window (temporal bootstrap, not i.i.d.).
Cyclic cross-temporal validation?
30 independent OOS windows; robustness = Sharpe distribution across splits, p5 percentile.
Fee model in batch5 MC?
0.10% entry + 0.10% exit (0.20% round-trip) — batch5_mc_optimizer.py.
How many MC params files on the VPS?
16 *_mc_params.json / *_v2_params.json files in /data/.
Does Monte Carlo guarantee real profits?
No — public paper trading; MC filters historical robustness, not a return promise.