A backtest that fills every order at the last printed price will make almost anything look profitable. Before you put capital behind a strategy, you need an honest picture of how it actually fills, and which costs your backtest does and doesn’t model. Here’s what to evaluate.
Why pre-live fill checks matter
The difference between a backtest and reality is mostly in the fills. Get the fills wrong and the equity curve is fiction. So the pre-live question isn’t “what return did the backtest show?”, it’s “how realistic were the fills that produced it?”
Slippage: model it, then measure it
Slippage is the gap between the price you expected and the price you got. Two steps:
- Model it in the backtest. Fills should be adjusted for an estimated spread and slippage rather than assuming the mid or last price. A realistic model ties this to the instrument’s typical range and tick size.
- Measure it in paper and live. Compare actual fill prices to what the strategy assumed. A consistent gap means your slippage assumption needs tuning; a wild, random gap points to a liquidity or timing problem.
Order types: know what your backtest models
This is where assumptions quietly bite. Not every order type behaves the same way in a backtest as it does at a broker.
- Market and limit orders are the bread and butter and are typically what a backtest fill engine models directly.
- Stops and stop-limits are trickier. A backtest may not fill them the way a real broker would, don’t assume a backtested stop reflects live behavior. If your strategy depends on stops, validate that behavior carefully in paper before trusting it.
Know which order types your strategy relies on, and treat any type your backtest doesn’t faithfully model as an open question, not a settled one.
Costs beyond the fill price
Slippage isn’t the only cost. Commissions and fees also matter, and how they’re accounted for depends on your broker. Don’t assume a backtest includes broker-specific commissions unless you’ve explicitly modeled them, in live, those fees come from the broker and should be reconciled against your records.
Order state monitoring
Before and during live, watch the order states: working, filled, partially filled, cancelled, rejected, expired. Rejections and unexpected cancels are early warnings that something, buying power, a bad parameter, a venue rule, is off.
A pre-live checklist
- Are fills modeled with spread and slippage, not at an ideal price?
- Do measured paper fills line up with the backtest’s assumptions?
- Does the strategy depend on order types the backtest doesn’t fully model (e.g. stops)?
- Are commissions/fees accounted for the way your broker charges them?
- Are you watching rejections and cancels, not just fills?
How Periscøpe helps
Periscøpe’s backtest fill engine models spread and slippage from each instrument’s recent range, and every run records the full order lifecycle and the fills behind each trade, so you can compare assumed prices to actual fills across backtest, paper, and live. It’s also honest about its limits: the backtest models market and limit orders, so order types like stops should be validated in paper rather than assumed from a backtest. The goal is a fill picture you can trust before, not after, you commit capital.