ARISAI QUANT · ROTATION RESEARCH
Systematic crypto
rotation systems.
Four rules-based portfolio models. Independent selection logic, visible defensive states, and historical performance presented with the assumptions behind it.
THE MODEL ARCHITECTURE
Three independent models. One adaptive ensemble.
Each model has its own allocation and risk process. Simurgh combines the three as an additional portfolio framework rather than replacing them.
Shahin
Relative-strength selection with independent exposure and defensive-state controls.
Independent modelDarya
Separate ranking and confirmation engines, with selective risk and cash exposure.
Independent modelCyrus
Rules-based asset qualification, position controls and defensive allocation states.
Independent modelSimurgh
Blends Shahin, Darya and Cyrus using prior-only model history, capped influence and turnover-aware rebalance rules.
21D window · 50% source cap · 10% turnover thresholdHISTORICAL RESEARCH · REFERENCE MODELS
Compare results, not just return.
These are the complete Shahin, Darya and Cyrus reference-model statistics, plus the Simurgh ensemble · not the individual engine results.
Swipe horizontally to see all comparison columns →
| Model | Backtest equity multiple | CAGR | Sharpe | Sortino | Calmar | Max drawdown | Capture date |
|---|---|---|---|---|---|---|---|
| Shahin | 126.59× | 256.5% | 2.27 | 4.54 | 10.88 | −23.6% | 20 Oct 2026 |
| Darya | 517.70× | 416.1% | 2.42 | 4.69 | 14.56 | −28.6% | 20 Oct 2026 |
| Cyrus | 212.06× | 309.5% | 2.49 | 2.96 | 10.61 | −29.2% | 20 Oct 2026 |
| Simurgh | 325.24× | 358.2% | 2.80 | 5.50 | 16.80 | −21.3% | 20 Oct 2026 |
Source: dashboard screenshots supplied 20 Oct 2026. Historical simulated reference-model metrics, not realized investor returns. Figures are snapshot values and not live-updating; verify model version, backtest window, and stated transaction-cost assumptions before publication.
A BASELINE MATTERS
Test the ensemble against a simpler allocation.
The same Simurgh capture includes an equal 33/33/33 source-weight comparison. Simurgh shows a higher equity multiple in that snapshot, but the equal-weight comparison has a slightly better Sharpe and smaller maximum drawdown.
Simurgh Sharpe 2.80, Sortino 5.50, DD −21.3% · Equal Sharpe 2.81, Sortino 5.63, DD −19.6%. No claim of superiority across all metrics.
RESEARCH & METHODOLOGY
A backtest is a starting point, not proof of an edge.
Results deserve context. Here are the controls and questions that matter when evaluating a rotation model.
Point-in-time decisions
Completed daily bars, prior-only inputs and separation of signal time from modeled execution.
Execution & turnover
Next-open fill assumptions, transaction costs, rebalance thresholds and defensive allocations.
Benchmarks & failure tests
Compare buy-and-hold, simpler blends, drawdowns, sensitivity and out-of-sample evidence where available.
Read the research and evidence limitations
The published figures are historical backtests. They are not live realized returns, a promise of future performance, or a substitute for walk-forward and independent out-of-sample confirmation.
Where research reports are available, assess source histories and publication timing, leakage controls, alternative benchmarks, fee/slippage sensitivity, model variants, concentration, turnover and adverse regimes. A favorable result on one metric does not establish a durable executable edge.
Simurgh’s supplied dashboard specifies a strict next-open simulation with 5 bps modeled fees plus 5 bps slippage. The three source systems may have distinct backtest and execution contracts; their costs and samples should not be assumed identical without verification.
EXPLORE THE ROTATION SUITE
From research to a visible portfolio process.
Model states, current allocation targets and deeper research are available inside the AriSai Quant dashboard with Elite access. TradingView versions provide on-chart model context.
Use +/− to zoom · ←/→ to switch images · Esc to close · swipe on mobile while fitted