Max Drawdown Calculation

[Anagoge]

I was seeing some oddities with the max drawdown calculation, so I ran a test optimization using the built-in SampleMACrossOver strategy from 1990-today using day bars on the SPX index. For the MA parameters 26/280 (20;40;1, 200;400;10), for example, the max drawdown in the optimizer results grid is reported as 5.97%, but the graph shows a completely different number at 22%. Am I missing something here, or is this a bug? It looks like it is showing the most recent drawdown, not the max drawdown, in some (but not all) cases.

Some other results rows for this same optimization run have a matching grid and graph (see screenshot 2). I'm using 6.5.1000.7. This issue isn't specific to SPX, since it also happens with IBM, for example.

[NinjaTrader_JoshP]

Thank you for bringing this to our attention. We will look into it.

[NinjaTrader_Dierk]

Here is the issue:
- the graph displays the drawdown etc. at the end of the day
- whereas the summary tab show you the absolute max drawdawn

Since the max drawdown might have occurred in the midst of the day you won't actually see it on the graph, since the graphs are end-of-period reports.

[Anagoge]

Here is the issue:
- the graph displays the drawdown etc. at the end of the day
- whereas the summary tab show you the absolute max drawdawn
Since the max drawdown might have occurred in the midst of the day you won't actually see it on the graph, since the graphs are end-of-period reports.

I'm still confused. The max intraday drawdown (grid) should never be smaller than the end of day drawdown (graph) (though they could be equal), but I think my screenshot shows that the numbers in grid can be much smaller than the graph. The first screenshot shows ~6% in the grid and ~23% in the graph. How can the max intraday drawdown be 6% and the max end of day drawdown be 23%?

Or are you saying that one drawdown (the grid?) is the max loss in a single day/period and the other (graph) is the max loss over the entire backtest? If so, I'd label them that way to make that clear:

- Max Drawdown (Total)
- Max Drawdown (Day|Period)

[NinjaTrader_Dierk]

>> The max intraday drawdown (grid) should never be smaller than the end of day drawdown (graph)
Correct, that is a different bug pending which we will address for NT7. However even as this bug will be fixed, the max DD on graph in most situations won't match the max (intraday) DD on the grid.

[Anagoge]

Personally, I don't mind that the graph and grid might show different numbers in certain cases, though it might be worth documenting. For me at least, I'm only trading once per day (market on close), so those two numbers will be the same.

The issue where the grid shows the wrong drawdown number is more critical, in my opinion, since I was previously exporting the results grid of an optimizer run to Excel and then using the drawdown numbers to rule out certain parameter sets with large drawdown, but I wasn't making my decisions on correct data, until I went and manually corrected the drawdown data exported to Excel. That said, I'm a software developer myself, so I understand if for some reason it isn't reasonable to fix until NT 7.

[NinjaTrader_Dierk]

Thanks for your feedback.

[rcatani]

Max Drawdown in the Grid accurate?

Has this issue solved in NT7?
How do we have to consider Max Drawdown in the Grid? Intraday or at Closing bars?
Is it accurate?
You know this is one of the most important data to decision about using ir discarding a Strategy.

Than you in advance for your (I hope prompt) reply.

Roberto

[NinjaTrader_PatrickH]

Hello Roberto,

Thank you for your post and welcome to the NinjaTrader Support Forum!

This has been updated for NinjaTrader 7. I can confirm the Max Draw Down will match the max level of Draw Down displayed on the Graphs tab of the Strategy Analyzer results. Please refer to the attached screenshot for an example.

Please let me know if I may be of further assistance.