A regular backtest can change when its available historical dataset or starting point changes, and repainting strategy logic can also create differences after reload. Compare the exact dataset and strategy settings before judging the result.
What you may be seeing
- Yesterday's backtest numbers do not exactly match today's
- Trades move or disappear after refreshing the chart
- Net profit or trade count changes even though you did not intentionally optimize the strategy
Most likely causes
- The regular chart dataset starts at a different historical point
- The amount of accessible historical bars changed
- The strategy uses logic that behaves differently on realtime and historical bars
- A strategy input, property, timeframe, or script version changed
Try this first
- Record the symbol, timeframe, script version, inputs, strategy properties, and date range
- Export Strategy Report results when you need a durable historical baseline
- Compare the amount and starting point of chart history
- Review the strategy for repainting if live signals change after bars become historical
If that didn’t fix it
- Use a controlled fixed testing window where possible
- Use Deep Backtesting when appropriate and available if you need a broader historical dataset
- Keep versioned copies of the Pine strategy when you change code
Verify the fix
You can explain the changed result by a dataset, settings, script-version, or repainting difference and can reproduce a stable comparison with those variables controlled.
Why this happens
Regular TradingView strategy results can change when the historical dataset available to the chart changes. TradingView explains that the starting point of chart history can move over time, and different plans can also provide different amounts of historical bars.
A repainting strategy can create another source of differences because realtime and historical calculations may not behave identically. Before assuming the backtester is broken, compare the exact symbol, timeframe, strategy version, inputs, properties, available history, and whether the strategy uses repainting logic.