Why doesn't my backtest match a paper trade with the same configuration?

Though in many cases running a Backtest with the same configuration over the elapsed time of a Paper Trade should be quite similar, there will generally be some deviation in results, sometimes more than expected. This is because Paper Trades are executed over sub-minute candle data, whereas the finest resolution a Backtest can achieve is one minute.

The deviation between Paper Trades and Backtests will tend to be much larger when the Buy/Sell Action Time is set to Realtime, and/or when Stop Loss/Take Profit are enabled. This is because Paper Trades can react quicker to price changes, and depending on the amount of fluctuation within a minute the results can vary. For example, if during an open position a price drops by 20% but recovers to only a 10% drop by the end of the minute, a Paper Trade may trigger the Stop Loss at a 20% loss whereas the Paper Trade would sell at a 10% loss. These differences can cascade throughout the lifespan of a simulation.

As a rule, you should expect Paper Trades to be more accurate than Backtests when comparing to a Live Run (though there will still be deviation as Live Runs have more variables due to running on a live exchange - ie. What is Slippage?).