Indicators in backtest vs reallife

[Uregon]

Is there a solution that the indicators work same as in backtest for the live market ?

[NinjaTrader_Jesse]

Hello Uregon,

That's highly dependent on how the indicator is programmed and the settings you use in your strategy.

Backtesting runs OnBarClose where in realtime you also have OnEachTick and OnPriceChange available, if the indicator uses those modes or if your strategy uses those modes then it won't work the same in a backtest.

This could also relate to what data the indicator needs, for example if it needs OnMarketData to be used then you would need to enable tick replay and make a set of logic specific for tick replay so it could be used there.

[siberfutures]

Hi, I like to address the backtesting issues that I encountered as well. Previously, I have version 8.0.25 and I recently installed 8.0.27 on a new computer. I used the same strategy, same indicators, same conditions and instruments. But I am getting huge difference in results. Example: 8.0.25 shows profit of $10k, while 8.0.27 shows loss of $10k on MES 12-22 and 03-23 has a slight difference.

Can you advise what was improved in 8.0.27 that is causing this and which version of backtest should we rely on for higher accuracy to be safe?
I understand live markets trading can be different from backtest, but it can't be that far off. I am using onbarclose for strategy and indicators as well. On another strategy, I used on price change because of trailing stop while indicators are barclose and it produces huge difference in results as well in both versions.

Thanks!

[NinjaTrader_Jesse]

Hello siberfutures,

You can use the change log to see what may have changed. Without knowing what specifically was affected in your strategies results it would be impossible to say what may be causing a difference. That version also is no longer available so there wouldn't even be a way to re test the situation.

https://ninjatrader.com/support/helpGuides/nt8/release_notes.htm

[siberfutures]

Hi Jesse, I read the updates release notes. It could be due to improved engine for historical data. Like I mentioned, everything is the same except for version 8.0.27.
No worries, I just have to retest everything from scratch as I couldn't count on the old backtest results anymore.
Also, after I updated on my old computer 8.0.25 to 8.0.27 without uninstalling the old version, the backtest results is the same as previous version even though it is now 8.0.27.
I deleted the cache to verify, but it remains the same results as before upgrade on the old computer.
Only on the new computer the backtest results are different.