How Tick-Based Backtesting Improves on Bar-Based Simulation
Summary
The document describes how a backtest advances through historical data and compares bar-based simulation with two tick-based approaches. In a traditional on-bar test, the strategy receives one decision point per candle, typically using its close, while intrabar price order and opportunities to trade during the candle are unavailable. This method is fast and straightforward but can miss intrabar behavior.
A simulated tick mode derives more frequent price updates from smaller underlying candles, creating additional decision points within each strategy candle and allowing entries and exits during that interval. A real-time tick mode uses recorded tick data at finer intervals, offering greater detail at the cost of slower runs and shorter practical test periods. The article illustrates the difference with a simple data-printing example and describes order matching against bid and ask prices. It cautions that even detailed replay lacks trade history, changing market depth, and actual network latency, so a backtest cannot fully reproduce live execution. The platform also provides features such as simulated network errors and delays for robustness checks.
Key ideas
- On-bar backtests provide one decision point per candle and cannot reveal the sequence of intrabar prices.
- Simulated tick replay creates more decision points by deriving price updates from smaller candles.
- Recorded tick replay offers finer historical detail but requires more data and runs more slowly.
- A more granular simulation can model intrabar orders, but still omits live depth changes, trade history, and network latency.
- Test results depend on the chosen data granularity and execution assumptions.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.