TradingView Backtesting Guide 2025: How to Test Any Strategy with Pine Script

Why Backtesting on TradingView Is Different

TradingView's Strategy Tester runs your Pine Script against historical OHLCV data bar by bar — exactly as it would execute in real time. This is bar-by-bar simulation, not vectorized backtesting, which means it catches intra-bar order execution and avoids certain look-ahead biases.

However, TradingView backtests execute orders at the next bar's open by default, not at the signal bar's close. This 1-bar delay models realistic execution. If you need signal-bar execution, use process_orders_on_close=true — but only if your broker guarantees fill at the closing price.

The Key Metrics in TradingView Strategy Tester

TradingView shows a large number of statistics. Here are the ones that actually matter and what to benchmark against:

• Net Profit — total P&L after commissions. Must be positive, but don't optimize for this alone
• Profit Factor — gross profit ÷ gross loss. Good: > 1.5. Excellent: > 2.0
• Sharpe Ratio — risk-adjusted return. Aim for > 1.0. Below 0.5 is poor
• Max Drawdown — largest peak-to-trough equity drop. Keep below 20% for most strategies
• Win Rate — % of trades that were profitable. Trend strategies: 35-50%. Mean reversion: 55-70%
• Average Trade — net profit per trade. Must cover commissions and slippage comfortably
• Total Trades — minimum 100 trades for statistical significance on any timeframe

Setting Up a Realistic Backtest

Most beginners run backtests with zero commission and no slippage — producing unrealistically good results. A realistic setup subtracts real-world costs before you evaluate performance.

strategy(
  title            = "Realistic Backtest Setup",
  overlay          = true,
  initial_capital  = 10000,
  default_qty_type = strategy.percent_of_equity,
  default_qty_value = 5,            // 5% risk per trade
  commission_type  = strategy.commission.percent,
  commission_value = 0.05,          // 0.05% per side (like most crypto exchanges)
  slippage         = 2              // 2 ticks slippage per trade
)

The Biggest Backtesting Mistake: Curve Fitting

Curve fitting (also called overfitting) is optimizing your strategy's parameters so heavily on historical data that it loses its edge on new data. Signs of curve fitting: very high win rate (>75%), perfect equity curve, parameters that look suspiciously specific (e.g., EMA(37) instead of EMA(20)).

The correct approach: split your data into in-sample (IS) and out-of-sample (OOS) periods. Optimize parameters on IS data only. Test the result without any changes on OOS data. If performance degrades dramatically, the strategy is curve-fitted.

A rough guideline: use 70% of available data for IS optimization, 30% for OOS validation. If the strategy holds up on OOS with at least 60-70% of IS performance, it has a genuine edge.

Walk-Forward Testing in TradingView

Walk-forward testing is the gold standard for strategy validation. You roll a fixed IS window forward in time, re-optimize at each step, and use the OOS period immediately after. Repeating this across the full data range shows whether the strategy's edge is consistent or regime-dependent.

TradingView doesn't have built-in walk-forward testing, but you can simulate it manually by changing the Strategy Tester date range. Set your IS period, note the optimal parameters, then shift the range forward and test with those parameters.

Tools like TradePilot can help you rapidly generate and iterate on backtest scripts so you can test more variants in less time.