What Backtesting Actually Tests (And What It Doesn't)
A backtest simulates how a defined ruleset would have performed against historical price data. Run correctly, it reveals whether your entry logic, exit logic, and risk parameters would have produced a positive expectation against real market conditions. Run carelessly, it produces a ruleset that looks profitable in hindsight, which is far easier than building one that works forward.
The honest limitations are as important as the capabilities. Backtesting cannot account for slippage on large orders in thin markets, cannot model the moments your psychology overrides a defined rule, and cannot predict regime changes. A strategy that performed brilliantly in the 2020–2021 bull market tells you very little about its behaviour in a prolonged bear. Treat backtest results as a filter, not a forecast. See a complete guide to backtesting trading strategies for the broader framework before treating any single result as a green light.
Choosing Your Strategy: SMA, RSI, MACD, Bollinger Bands, or Momentum
Each strategy type has a distinct character and performs differently depending on market conditions. Choosing one before configuring parameters prevents the confusion of optimising multiple approaches simultaneously.
An SMA crossover generates entries when a shorter moving average crosses above a longer one. It is a trend-following system that performs well in strong directional moves and produces frequent false signals in ranging markets.
RSI-based strategies enter when the asset is oversold and exit when it reaches overbought territory (a mean-reversion approach that works best when price oscillates within a defined range).
MACD strategies use the relationship between two exponential moving averages and a signal line. The histogram crossing zero provides entry triggers with slightly less lag than pure SMA approaches.
Bollinger Bands strategies enter when price touches or breaches the outer bands and exit on reversion toward the middle. The entry direction determines whether the strategy is breakout or mean-reversion in character.
Pick one, test it thoroughly, and resist the temptation to blend multiple indicators before you understand what any single one is measuring.
Configuring Parameters Without Overfitting to Historical Data
Overfitting is the most dangerous failure mode in backtesting. It happens when you adjust period lengths, thresholds, and filters until results on your test dataset look excellent. That is essentially memorising the past rather than finding an edge.
Parameter simplicity is the primary defence. Start with industry-standard defaults: for SMA crossovers, the 50-and-200 combination; for RSI, a 14-period length with 30/70 thresholds. Run the backtest on defaults first. Only after you understand what the defaults produce should you consider adjustments, and then change one variable at a time against a dataset you have not previously examined.
Walk-forward testing is the more rigorous approach: divide your historical data into an optimisation window and a validation window. Tune parameters on the first; test the result on the second without further adjustment. Performance that collapses on the validation window signals overfitting.
Reading the Results: Return, Win Rate, Drawdown, and Sharpe Ratio
Four metrics in the results panel each tell you something distinct about strategy quality.
Total return is the headline number and the most misleading. A strategy that produced 200% but required surviving a 70% peak-to-trough drawdown is not viable for most traders — holding through that drawdown in a real environment is practically beyond reach.
Win rate only becomes meaningful alongside the average win-to-loss ratio. A 40% win rate with winners averaging three times the size of losers is highly profitable; a 60% win rate with the reverse ratio is not.
Maximum drawdown reveals the worst realistic experience you would have had running the strategy live. Ask honestly whether you would have continued following the rules through that period.
The Sharpe ratio normalises return against volatility. Above 1.0 indicates meaningful return per unit of risk. Below 0.5, the volatility is not being adequately compensated. A high-Sharpe strategy is generally more deployable than a high-return, high-volatility one.
The Equity Curve: How to Spot Strategies That Only Worked in One Regime
The equity curve plots account value over time and reveals the complete story of how a strategy behaved across different market conditions. A healthy curve rises relatively consistently, with proportionate and recoverable drawdowns.
Watch for the pattern where the curve is flat for eighteen months and then spikes dramatically during one specific bull phase. That is a regime-specific bet, not a robust edge — the aggregate returns look good only because of one concentrated period.
Examine specifically how the strategy performed around major inflection points: the COVID crash of 2020, the late-2021 peak, the 2022 capitulation. A strategy that navigated these structurally different periods without catastrophic drawdown has demonstrated genuine resilience. One that collapsed in a specific regime has at minimum told you when not to deploy it.
Moving From Backtest Confidence to Live Paper Trading
A positive backtest earns a strategy a trial on a live paper account — not real capital. Paper trading is where you discover whether the theoretical edge survives contact with real spreads, real signal timing relative to candle closes, and the friction of watching positions move against you before resolving.
Run the strategy on the paper account with the exact parameters the backtest used. Track whether real-time results resemble the backtested metrics over at least thirty trades. Significant divergence between the two is diagnostic: it tells you where the simulation's assumptions parted ways from market reality, and that gap needs to be closed before any real capital is involved.
