Backtests
Historical simulation
Multi-year research curves for the paper-book target mix. This is not the Live Paper desk. Live results will differ.
Window · 01 Jan 2018 → 17 Jul 2026
Updated 17 Jul 2026 · historical simulation
Paper book (Alpaca 75% + FX 25%)
Alpaca sleeve mix at 75% of capital; FX desk 25%.
Full sample
- Sharpe
- 1.514
- CAGR
- 19.22%
- Vol
- 11.65%
- Max DD
- -15.53%
Out of sample (from 2024)
- Sharpe
- 1.189
- CAGR
- 12.28%
- Vol
- 10.08%
- Max DD
- -11.05%
Equity curve
Weekly marks, indexed from 100,000. Toggle combined book vs Alpaca-only.
Alpaca sleeves only
Full sample
- Sharpe
- 1.455
- CAGR
- 23.98%
- Vol
- 15.04%
- Max DD
- -25.05%
Out of sample (from 2024)
- Sharpe
- 1.265
- CAGR
- 17.00%
- Vol
- 12.98%
- Max DD
- -12.28%
Sleeves
Full-sample and out-of-sample metrics per sleeve. Allocations match the current paper-book targets. See Strategies for plain-English blurbs.
| Sleeve | Alloc | Sharpe | CAGR | Max DD | Out-of-sample Sharpe |
|---|---|---|---|---|---|
PEAD alpaca | 30%of Alpaca | 1.003 | 21.5% | -33.8% | 1.146 |
Crypto breakout alpaca | 20%of Alpaca | 0.862 | 26.6% | -49.4% | 0.061 |
VIX reversion alpaca | 20%of Alpaca | 1.247 | 25.9% | -26.4% | 1.497 |
Equity momentum alpaca | 12%of Alpaca | 0.909 | 24.6% | -42.9% | 1.313 |
Risk parity alpaca | 10%of Alpaca | 0.670 | 5.3% | -19.1% | 1.319 |
Crypto vol core alpaca | 5%of Alpaca | 0.664 | 16.7% | -39.9% | 0.265 |
Crypto BTC relative alpaca | 3%of Alpaca | 0.550 | 15.6% | -57.6% | 0.525 |
FX macro desk oanda | 25%of total | 0.379 | 4.3% | -22.1% | -0.105 |
Crypto momentum alpaca · parked | 0%of Alpaca | 0.665 | 25.9% | -81.2% | -0.093 |
Methodology
Daily sleeve returns from the same engines that run the paper book. The combined curve weights the Alpaca sleeve mix at 75% of total paper capital and the FX desk at 25%. Transaction costs are included in each sleeve. Out-of-sample metrics use 2024 onward. Equity curves are weekly and start at 100,000.
Closer to realism
- Next-day P&L on prior weights (no same-bar free lunch)
- Costs scale with turnover; crypto heavier than equities
- Leverage and size caps; rebalance frequency matches sleeve
- Book built from the same sleeve engines and current target weights (75% Alpaca / 25% FX)
- Research filters (out-of-sample, correlation, max drawdown). The published book is not every in-sample curve
Still not fully realistic
- Daily close-to-close bars; no intraday path or partial fills
- Slippage is a flat percent. Real impact rises with size and illiquidity
- No borrow / short locate (fine for this long-biased book)
- Crypto weekends and FX alignment are approximations
- Paper is not live: optimistic fills; no halts, fat-fingers, or API downtime
- Survivorship and data vendors are imperfect