Backtest Results 2022-2025
1,869 anonymized halal-only spot trades across three tiers. Calibrated to whitepaper §7. Free under CC-BY-4.0.
CSV
486 KB · 1,869 rows
Download →
Parquet
163 KB · snappy
Download →
JSON API
filterable · CORS-open
Open endpoint →
Tier results
Cumulative returns are calibrated to whitepaper §7. Win-rate and max drawdown are reported as observed in the seeded dataset (within ±5% of the whitepaper figure).
| Tier | Trades | Win rate | Cumulative | Max DD |
|---|---|---|---|---|
| Conservative | 376 | 60.1% | +47.3% | -6.4% |
| Moderate | 436 | 58.9% | +118.4% | -9.5% |
| Multi-X | 1,057 | 42.3% | +214.7% | -23.3% |
Schema
| Field | Type | Description |
|---|---|---|
| trade_id | string | Stable per-tier identifier (e.g., CON-00042). |
| tier | enum | Conservative · Moderate · Multi-X. |
| timestamp_utc | ISO 8601 | Trade close time in UTC. |
| pair | string | Spot pair (e.g., BTC/USDT). Halal-only universe. |
| side | enum | Always BUY — spot-only buy-then-sell round trips. |
| entry_price_usd | float | Entry mid-price in USDT. |
| exit_price_usd | float | Exit mid-price in USDT. |
| holding_period_minutes | int | Time between entry and exit. |
| size_pct_of_equity | float (0-1) | Position size as fraction of equity. |
| notional_unit_portfolio | float | Notional in unit-portfolio terms (start = 1.0). |
| position_pct_return | float | Return on the position itself (entry → exit). |
| pnl_pct | float | Portfolio-level contribution. Cumulative compound matches whitepaper. |
| pnl_unit_portfolio | float | Realized P&L in unit-portfolio terms. |
| decision_rationale | enum | Tagged reason from the agent chain. |
| agent_chain | string | Pipeline of agents that fired (→-separated). |
| halal_status_at_trade | enum | Asset's halal classification at trade time. Always 'halal' (only halal assets are traded). |
| exchange_venue | enum | 'spot' — derivatives/margin/perp are excluded by design. |
| is_anonymized | bool | Always true. No customer identifiers. |
API examples
# All trades
GET /api/datasets/backtests-2025-2026
# Filter by tier
GET /api/datasets/backtests-2025-2026?tier=Multi-X
# Filter by pair + date window
GET /api/datasets/backtests-2025-2026?pair=BTC/USDT&from=2024-01-01&to=2024-12-31
# Limit + offset for pagination (default limit 200, max 5000)
GET /api/datasets/backtests-2025-2026?limit=50&offset=100
# Programmatic access (Python)
import pandas as pd
df = pd.read_parquet("https://gethalalcrypto.com/datasets/backtests-2025-2026.parquet")
df.groupby("tier")["pnl_pct"].agg(["count", "mean", "sum"])How to cite
HalalCrypto Research. (2026). HalalCrypto Backtest Results Dataset 2022-2025 (v1.0.0). Anonymized synthetic-trade dataset calibrated to whitepaper §7 cumulative returns. CC-BY-4.0. Retrieved from https://gethalalcrypto.com/datasets/backtests-2025-2026
Methodology
Per-tier outcomes are sampled from a gamma distribution parameterized by the tier's configured win-rate, average winner, and average loser. Multiplicative scaling anchors the cumulative compounded return to the whitepaper figure exactly. Pair allocation reflects the tier's typical universe (more BTC/ETH at Conservative; broader alt mix at Multi-X). Timestamps are uniformly distributed over 2022-2025.
Limitations
This is a synthetic, anonymized dataset — not a live trade log. Use it to study tier shape, agent rationale distribution, and per-pair concentration; do not treat individual rows as historical fact. Live performance reporting (when available) will be published as a separate dataset versioned alongside this one.
Companion paper
HalalCrypto Whitepaper 2026
36 pages · methodology + governance
Sibling dataset
Top-200 halal-coin classification
CSV + JSON · AAOIFI-aligned screening
Reference
Halal methodology
Three-gate pipeline · AAOIFI SS-21