Improving Backtest Realism with Costs and Intrabar Data
Summary
The article argues that a backtest should approximate live trading conditions rather than maximize the appearance of historical returns. It recommends including commissions and slippage, with estimates adjusted to the instrument and checked against actual execution records. Its examples emphasize that costs can vary sharply across markets and liquidity conditions, and can materially change a strategy’s results.
It also explains how bar-based simulations can misrepresent trailing stops: OHLC data may hide price movements within a bar that would have triggered an exit in real time. Suggested remedies include intrabar testing, using a finer-resolution dataset for stop tracking, or substituting a conservative fixed target when finer data are unavailable. The closing checklist flags same-bar entries and exits, overnight exposure, small samples, illiquid securities, leveraged products, and overlapping symbols as potential sources of misleading results. These are practical diagnostics rather than proof that any strategy will work; the article offers no controlled performance study, and its cost examples are specific to the author’s experience.
Key ideas
- A useful backtest models commissions and slippage using instrument-specific evidence where possible.
- Bar-level data can conceal intrabar price paths that would trigger trailing stops in live trading.
- Finer-resolution data can improve stop simulation, while conservative fixed targets offer a rough fallback.
- Small samples, illiquid assets, overnight exposure, and overlapping instruments can distort backtest results.
- Backtests validate rules under assumptions and should not be judged only by the shape of an equity curve.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.