Methodology
How we build, test, and validate every strategy on Loot-Fi.
Data
All strategies are backtested on historical Binance Futures (USDT-margined perpetual) daily data. We use OHLCV candles and 8-hourly funding rate snapshots. The dataset currently covers 500+ instruments dating back to 2020.
Funding rate strategies use Binance's native 8-hour funding intervals, aggregated to daily where required. Scanner data also incorporates Hyperliquid perpetuals for broader market coverage.
Universe filters
Not every coin is tradeable. We apply volume and history filters to ensure strategies only trade liquid instruments with sufficient data:
| Filter | Value |
|---|---|
| Minimum daily volume | $500K - $10M (varies by strategy) |
| Minimum history | 90 days |
| Volume lookback | 60-day rolling median |
Each strategy page states its exact volume threshold. Funding rate strategies use a stricter $10M filter since they require deep liquidity for both legs.
Fee assumptions
Every backtest includes realistic trading costs:
| Cost type | Assumption |
|---|---|
| Round-trip trading cost | 7 bps (3.5 bps per side) |
| Slippage | Included in fee estimate |
| Funding rate costs | Deducted from funding strategies |
The 7 bps round-trip assumption reflects taker fees on major exchanges. Strategies that rebalance frequently pay more total costs; this is fully accounted for in all reported metrics.
Position sizing
Most strategies use volatility-targeted position sizing. The framework calculates realised volatility over a trailing window and scales exposure so the portfolio targets a specific annualised volatility level.
| Parameter | Typical value |
|---|---|
| Vol target | 40% annualised (varies by strategy) |
| Vol lookback | 60-day rolling |
| Max leverage | Capped per strategy (1x - 5x) |
| Rebalance frequency | Daily or weekly (stated per strategy) |
Performance metrics
All metrics are calculated on daily returns after fees:
- Sharpe ratio: annualised (daily mean / daily std * sqrt(365))
- CAGR: compound annual growth rate from equity curve
- Max drawdown: peak-to-trough decline in equity
- Win rate: percentage of positive daily returns
- Annualised volatility: daily std * sqrt(365)
Yearly breakdowns show return, Sharpe, and max drawdown per calendar year. Monthly heatmaps show returns by month.
Robustness testing
Every strategy undergoes up to 7 robustness tests. A strategy must pass at least 4/7 to receive a positive robustness score:
- Sample size: Minimum number of trades for statistical significance
- Regime test: Performance in bull vs bear markets (BTC above/below 200-day MA)
- Decay test: Whether alpha degrades over time (comparing first vs second half)
- Monte Carlo: Bootstrapped return distributions to assess consistency
- Walk-forward: Rolling out-of-sample windows to check for overfitting
- Parameter sensitivity: Performance across nearby parameter choices
- Fee sensitivity: Whether the strategy survives at 2x the assumed fee level
Alpha stacking
Alpha stacks combine multiple strategies weighted by allocation. We analyse correlation between component strategies and report blended portfolio metrics. Stacking benefits come from diversification across uncorrelated return streams:
- Funding rate strategies (market-neutral, low correlation to price)
- Momentum/breakout strategies (trend-following, directional)
- Mean reversion strategies (counter-trend)
Stack-level metrics (Sharpe, CAGR, MaxDD) are computed on the combined daily return series, not averaged from components.
What we do not do
Transparency about limitations is as important as methodology:
- We do not optimise parameters on the full sample and report those results as expected performance
- We do not backtest on illiquid coins that cannot realistically be traded
- We do not ignore trading costs or funding rate drag
- We do not cherry-pick time periods to inflate metrics
- We do not claim backtested results predict future returns
Limitations
All backtests are hypothetical. They assume fills at daily close prices, which may not be achievable in practice. Slippage on large orders, exchange downtime, and market microstructure effects are not modelled. Crypto markets are young and regime shifts are common. A strategy that worked historically may stop working without warning.
Nothing on Loot-Fi constitutes financial advice. Use this research as one input among many.