Strategy performance in Playback Historical vs Playback Real-Time

[2inthebush]

I've read everything I can find on this forum and the user guide about how Strategy Analyzer and Playback works, but it seems that I'm still confused because I can't explain the wild difference in strategy performance in Playback Historical versus Playback Real-Time (Historical). The historical data I'm using is bid/ask/last tick data from Continuum. I'm testing a very simple price action strategy, so I presume that Level II and volume data shouldn't affect the performance in Playback. All calculations are set to Calculate On Bar Close. Therefore, I don't test using Playback Market Replay. Maybe I'm using it incorrectly.

When I test a strategy in Playback Historical, first I open a chart with the instrument data, and set the Days to Load for x days prior to the days I will test with the Playback Historical control window. Then I add the the strategy to the chart, choose High Resolution Fill, and enable the strategy. Then I start the Playback Historical player, and allow it to complete Playback.

After Playback completes, I look at the strategy performance results in Historical and Real-Time. Playback Historical always results in vastly better strategy performance for every strategy I've tested. For example, a net profit of $10K in Historical, versus a net profit of $1K in Real-Time for the same time period. The Playback Historical results are based on the x days of charts loaded prior to the date range that I chose for the Playback Historical control window, whereas the Real-Time Playback Historical results are based on the date range that I played using the Playback control window.

I don't understand why Playback Historical always returns vastly better results. Aren't Playback Historical and Playback Real-Time (Historical) processing the same historical data with identical fill resolution? Why is there always a huge improvement in strategy performance in the Playback Historical results?

[NinjaTrader_ChrisL]

Hi, thanks for your post.

The Market Replay data contains Level 2 data where Historical repay does not. That could cause differences in trade calculation. You can also use the Print statement to print out times, price, and indicator data values when your trade condition becomes true to see where exactly the differences are coming from. Also see here for explications of some differences that can come up in comparing real time vs backtesting strategies:
https://ninjatrader.com/support/help...ime_vs_bac.htm

Best regards,
-ChrisL

[2inthebush]

After comparing the trades between Playback Historical and Playback Real-time, I believe I have figured out my issue. It seems that Playback Real-time is calculating more realistic slippage. In Strategy Analyzer I'd used the setting to add slippage. I'd presumed that Analyzer was subtracting the slippage from the net profit calculation. However, I now realize that I need to manually subtract the slippage results from the net profit results in Analyzer This would explain the huge drop in strategy performance when testing in Playback Real-time.