Forward testing: the stage after the backtest

01Why forward test at all

A backtest uses history that the strategy's author has seen. A forward test uses data that did not exist when the rules were written, so it cannot be overfitted to it. It also reveals practical issues a backtest hides: orders rejected, partial fills, data feed gaps, platform restarts, session times set wrongly, and behaviour that differs from the simulation.

02Stage one: simulation (paper trading)

NinjaTrader provides a simulated trading account that works with live market data, so the strategy places simulated orders in real time. Run it long enough to see a range of conditions and a meaningful number of trades. Things to confirm:

• The strategy takes the trades the backtest says it should take on the same data.
• Stops and targets are placed correctly and cancel each other properly.
• It behaves sensibly at the session start, the end of day and around a disconnect.

Be aware that simulated fills are optimistic. A simulator may fill limit orders that a real market would not, and it does not feel your order's effect on the book. Treat sim results as a check of the mechanics, more than a check of profitability.

03Stage two: small size live

Only real money exposes real fills, commissions, slippage and your own reactions. Start with the smallest practical size and run for a set period or number of trades defined in advance. The aim is to learn how live results compare with the backtest and the sim, not to make money yet. If the account type is a prop firm evaluation, apply the drawdown and daily loss rules even in testing.

04What to measure

• Number of trades and how it compares with expectation.
• Average win, average loss and win rate against the backtest figures.
• Slippage per trade, which is the clearest live-versus-test gap.
• Drawdown and how it compares with your tested worst case.
• Rule compliance: did every trade follow the rules? Differences point to bugs or interference.
• Operational issues: disconnects, missed orders, restarts.

Keep a simple journal so differences are recorded, not remembered.

05When to stop a strategy

Decide the stopping rules before trading, when you are calm. Examples of reasonable triggers:

• Drawdown beyond the limit you set from your testing.
• Live behaviour that departs clearly from the rules, which suggests a bug.
• Persistently worse costs than your assumptions, enough to erase the edge.
• A change in the market or the instrument that breaks an assumption of the idea.

A normal losing streak is not a reason to stop, but a streak much worse than your tests suggested is a reason to look. Distinguishing the two is the reason to measure, and to keep your backtest records. Stopping is not failure; it is the system working. Past performance does not guarantee future results, and a strategy that worked for a period may stop working.

If you want to compare notes with other traders, there is a community on Discord, and our parent brand Trade With Excellence has more resources.

FAQFrequently asked questions