StrategiesREFERENCE
6 built-in strategy definitions · performance figures are measured by the backtest engine below, or omitted where the rules cannot be tested
Walk-forward, replicated on two asset classes — these are risk tools, not alpha
Five portfolio variants ranked on a tuning window, then the winner run once on data it never saw. Run separately on equities and on crypto — different decades, different volatility regimes, same outcome.
| Universe | Tuned | Tested | In-sample edge | Out-of-sample edge | Drawdown saved | Variants w/ edge |
|---|---|---|---|---|---|---|
| 10 US equities | 2006–2015 | 2016–2026 | +0.27 | -0.10 | +38.3pp | 2/5 |
| 7 crypto pairs | 2018–2022 | 2023–2026 | +0.09 | -0.21 | +16.7pp | 0/5 |
Both universes showed a positive edge in-sample and a negative one out-of-sample. On crypto, zero of five variants held any edge at all. What generalised in both, without exception, was drawdown reduction. Read every single-window backtest below as an artifact of its window, not as evidence of edge.
Backtest Engine
real OHLCV · costs included · runs in your browserA cold sweep is paced at ~8s per uncached equity symbol to stay inside Twelve Data's 8 requests/minute limit — roughly 90s the first time, then instant for 12h from cache.
Long while SMA50 > SMA200 and price holds above SMA50. Exit on death cross or an 8% stop.
Approximation — the published rules also require, and this does NOT model:
- · EPS growth ≥ 25% YoY (no free fundamentals feed)
- · RS Rating ≥ 85 vs S&P 500
- · VCP base geometry and pivot detection
- · Mutual fund accumulation weeks
Momentum Surge
Not backtestable
Measuring…
Mean Reversion
Not backtestable
Measuring…
News Catalyst
Not backtestable
Needs historical news + sentiment scores — no free archive
Copy Trade
Not backtestable
Needs point-in-time 13F / bulk-deal history
Cross-Market Arb
Not backtestable
Needs paired ADR + NSE ticks and intraday FX
Geo Hedge
Not backtestable
Needs a GDELT risk-score time series