Direct answer
Backtesting is the process of testing a trading idea against historical market data to see how the idea could have behaved if it had been applied in the past. In forex backtesting, the idea is usually converted into explicit rules (for example, when to enter, when to exit, and how much to trade). The purpose is evaluation and learning, not prediction.
Mechanism and definition
A practical backtest turns a concept into a repeatable checklist. Typical inputs and components include:
- Historical price series: the time-stamped data used for the test. In forex, data typically includes prices (and may be supplemented by assumptions about spreads or execution).
- Strategy rules: entry and exit conditions, position sizing method, and any holding-time logic. These rules must be unambiguous.
- Execution model: how trades are assumed to fill. A common simplification is assuming trades fill at a particular price, but real fills depend on bid/ask, slippage, latency, and liquidity.
- Costs: commissions, swap/financing effects, and trading costs. If costs are omitted or approximated, results can be overstated.
- Assumptions and settings: timeframe, leverage treatment (if modeled), risk limits, and how missing data is handled.
During the backtest, the rules are applied step-by-step through the historical data. The system records what trades would have occurred and calculates outcome metrics such as drawdowns, profitability measures, and trade frequency. The key point is that these metrics are functions of both the idea and the assumptions.
Evidence or example (with explicit assumptions)
Consider a simplified forex idea written as rules:
- Use only one currency pair.
- Every time condition A occurs, open a position at the next time step’s mid price.
- Close after a fixed number of bars (for example, after N bars).
- Use a constant position size.
- Ignore financing costs and assume no slippage.
If condition A occurs many times during the chosen historical window, the backtest will produce a sequence of hypothetical entries and exits, then aggregate results. However, this example demonstrates the main verification idea: if you change any assumption—such as filling at bid/ask instead of mid, adding spread, including swap/financing, or using different bar counts—the results can change materially.
Limitations and risks
Backtesting has material failure modes:
- Overfitting: When rules are adjusted repeatedly to match one historical period, they may capture noise rather than a reusable pattern.
- Regime change: Forex market behavior can shift (volatility, liquidity, macro conditions). Historical relationships may not persist.
- Unrealistic execution: Backtests often use simplified fill prices. Real trading can differ due to spread variation, slippage, and delayed fills.
- Data quality and alignment: Incorrect time alignment, look-ahead bias, or inconsistent data can make results look better than they would be in real time.
- Cost omissions: Ignoring commissions, spreads, and financing can inflate apparent performance.
Because of these limits, historical success is not evidence that the same rules will work in the future.
Verification or next question
To independently verify a backtest, focus on whether the process is checkable and falsifiable:
- State every assumption (prices used, fill assumptions, costs, and time window).
- Use out-of-sample testing: evaluate on a different time period than the one used to refine rules.
- Test robustness: vary parameters and see whether the outcome depends on a narrow setting.
- Check for look-ahead bias: ensure that only information available at the decision time is used.
A useful next question is: “Which specific assumptions, if changed, would most strongly alter the results?” If you cannot answer that, the backtest may not be reliably interpretable.