Backtesting crypto strategies is not just equity backtesting with different tickers. The market is open continuously, liquidity is fragmented, and venue risk can be as important as price risk.
The first design choice is time. Researchers must define the time zone, bar boundaries, and rebalance schedule. A daily bar ending at midnight UTC can produce different signals from a bar ending at another time. If the strategy trades every eight hours, the schedule should be documented and tested consistently.
The second choice is data source. Crypto price data can differ across exchanges because liquidity, fees, outages, and local demand differ. A robust backtest should identify whether it uses spot prices, futures prices, volume-weighted data, or a consolidated index.
The third issue is survivorship bias. Many crypto assets disappear, lose liquidity, or become untradable. A universe built only from today's surviving tokens can overstate historical performance. Delistings and inactive assets should not be silently removed when the research question depends on broad market exposure.
Execution assumptions need special care. Smaller tokens may show attractive historical moves but have spreads or order book depth that make real fills unrealistic. Strategies that rebalance frequently should include spread, fee, and slippage assumptions.
Operational risk is also part of crypto backtesting. Exchanges can halt withdrawals, APIs can fail, stablecoins can depeg, and funding rates can change quickly. A live system needs monitoring and fallback rules for these events.
A good crypto backtest is explicit about what it does not model. The goal is not to make the simulation perfect; the goal is to make the assumptions visible enough that readers can judge the evidence.