Why futures funding belongs in every crypto backtest
A futures strategy can look viable before funding and fail once recurring holding costs are included. The longer positions remain open, the more important the separation becomes.
Funding compounds with holding time
Count how many funding intervals a typical position crosses and whether the strategy tends to pay or receive funding. A small recurring debit can dominate a low-frequency edge.
Separate trading PnL, fees, and funding
A single net number hides the reason a strategy works or fails. Review the three components independently and compare them across holding periods.
Stress the rate assumption
Historical funding is not a promise about future rates. Test whether a less favorable rate or a longer holding period removes the expected advantage.
Audit one futures result
Use an existing report or the public demo and answer these questions.
- Record gross trading PnL, total fees, total funding, and average holding time separately.
- Identify which positions crossed funding timestamps and whether their direction usually paid or received.
- Recalculate the decision using a less favorable funding assumption and note whether it still passes.
Minimum research record
Keep these inputs with every result so the test can be reproduced and compared without relying on memory.
- Exchange, market type, symbols, timeframe, and exact date range.
- Starting balance, position sizing, leverage, and risk limits.
- Entry, exit, stop-loss, take-profit, and filter configuration.
- Fee, funding, and execution assumptions used by the simulation.
- Net result, drawdown, trade count, losing periods, and known anomalies.
What this evidence cannot prove
A backtest cannot guarantee future returns or fully reproduce liquidity, order-book depth, latency, slippage, rejected orders, exchange downtime, or future funding rates. Use the result to reject weak assumptions and design the next controlled test—not as permission to take unlimited risk.
Find out what actually produced the net result
Inspect the demo report with costs in view. A strategy should not advance because gross PnL hides an unfavorable funding profile.