Drawdown Question

[r2kTrader]

DrawDown Issues

Dierk,

http://www.ninjatrader-support2.com/vb/showthread.php?t=13316&goto=newpost

Josh was kind enough to address both my question and concerns and offered a constructive direction.

The above post is for anyone who was having similar issues/concerns that I was and was looking for an alternative direction to get the information pertaining to the max unrealized drawdown.


Regards,



r2kTrader

[PipSqueak]

Please explain NT drawdown calculations. I have a backtest run where NT shows a max drawdown of -9%. The max drawdown that I see, as shown below, is -50; the 1st column is cumulated gain. This is on closed positions, not open. Attached is a spreadsheet with the trades and the NT performance report.

Just how is NT calculating the drawdown and what is it supposed to signify? - I can't see how it comes up with 9% which is highly misleading. For every strategy test, you will have to manually determine the drawdown to know how far down it really goes. The NT DD number is quite far from reality and, well, unusable.

2975.0
2750.2 -8%
2513.3 -16%
2270.1 -24%
2038.3 -31%
1795.8 -40%
1717.8 -42%
1482.4 -50%

[NinjaTrader_JoshP]

PipSqueak,

Please see this article: http://www.ninjatrader-support.com/H...xDrawdown.html

Not sure you are using the right calculations on your end.
1717.8 to 1482.4 is not 50% drawdown for sure. 50% of 1717 is roughly 850.

[PipSqueak]

Going straight from 2975 to 1482 is a drawdown percentage of
(1 - 1482/2975) = 50.2%. This is the maximum drawdown in this backtest of the system, not 9% as indicated from the NT performance report.

What does "local" mean in that NT help link?

[NinjaTrader_JoshP]

Local means between each individual trade. It is not calculated from a global extrema like you have outlined in your calculation.

[MrSmartGuy]

I see this thread went inactive six years prior. I found it just today, when looking for insight on why the drawdown statistics don't appear to be working with my strategies (backtesting). I am experiencing exactly the issues that r2kTrader, and would like to know if the drawdown graphs are supposed to be working yet? Hard to imagine that this is by design -- indeed, if you sell hot dogs, then make the mustard available. Need something more than a suggestion to find a field of mustard, gather some mustard seed, and then grind it up. What am I missing?

[NinjaTrader_PatrickH]

Hello MrSmartGuy,

Thank you for your post.

The Drawdown should be working. Can you attach a screenshot of the results you are seeing? And are you testing Forex instruments by chance?

[MrSmartGuy]

Hello Patrick,

Attached is a screenshot of the chart and the performance summary box. Please note that no drawdown or MAE is shown for the backtest. The rightmost trade on the chart clearly has intra-trade drawdown occurring.

Thanks for looking into this,

Brad

[NinjaTrader_PatrickH]

Hello MrSmartGuy,

Thank you for your response.

The MAE is to small, you can see it is listed red but due to the fact NT 7 does not show the values past the second decimal place it just shows red.

For the Max Drawdown, are we looking at Historical, Historical and Real-Time, or Real-Time Performance for the strategy?

[MrSmartGuy]

I have attached a second screenshot file, identical to the first, except that performance values are now shown in currency rather than per cent. With an intra-trade drawdown of eight points for 500 shares of SPY in the most recent trade, shouldn't there be something other than zeros shown for the MAE? Also, the Max Drawdown still shows all zeroes as well.

The performance report was generated for "historical" results.

Thanks for reviewing,

Brad

[NinjaTrader_PatrickH]

Hello MrSmartGuy,

Thank you for your response.

On the Trades tab, do your trades show MAE for Currency or Percent?

[MrSmartGuy]

I have attached an additional screenshot, showing the Trades tab data. MAE values are zero for all trades, and shown in "currency."

I decided to break out my Sherlock Holmes super sleuth detective kit, whereupon I noticed that, in addition to zero MAEs for all 31 trades, there are also 31 non-zero MFE values. I deduced that this was more than mere chance, as this particular strategy has the rare, and unlikely real-life, situation of all trades being profitable.

Given this intriguing correlation, I further calculated that the MFE values are nothing more than the end-to-end profit of each trade, with no accounting for intra-trade movement. In this latest screenshot, I have shown a different trade (No. 21, highlighted in the trade list,) as it has a very clear upside move, well beyond the final exit price, but the trade list clearly shows that only the final exit price has been used.

I have asked Dr. Watson to review my findings. In the meantime, I ask that you also peruse them. Hopefully, at the end of the day, we will discover that only a simple errant Ninja has knocked us off of our horse, and that we are not entering battle with Professor Moriarty.

[NinjaTrader_PatrickH]

Hello MrSmartGuy,

Thank you for your response.

There is a known issue with the MAE based on connections actually. Are you connecting to multiple connections? If so, do you connect to your brokerage connection first or your data feed provider first?

[MrSmartGuy]

Hi Patrick,

At this time, I am running some backtest studies, and only connected to a data feed, with no broker tie-in.

It would help if you could tell me how the drawdown, MAE, and MFE statistics should work, along with any known anomalies of user configurations or software bugs that might gum up the works.

So, starting at the top, should drawdown, MAE, and MFE include intra-trade drawdown and excursions?

Thank You.

[NinjaTrader_PatrickH]

Hello MrSmartGuy,

Thank you for your response.

The values should be present even for historical backtesting. The MAE and ME should consider the closes of all bars in-between at the least (should also include the Highs and Lows). Which is clearly not occurring on your install.

Can you try repairing the database by going to Tools > Options > Data > Repair Database?