Rocket Science Mini Dow drawdown question

I noticed the max drawdown for Rocket Science Mini Dow is 11.61%, but the per-trade drawdown is 17.7% of equity for the trade closed on 1/31.



I expected the former could not be smaller than the latter. What am I missing here?

never mind, it’s probably b/c the trade had two entries.

The max draw down in the advanced statistics is based on daily equity values. I suppose these are the EOD values, so intraday drawdowns are not included in them. The drawdown per trade has a specific time attached to it, so I suppose it is also affected by intraday drawdowns.

Really? Then I’d suggest fixing this important basic stat now instead of adding tons of new advanced stats (based on incorrect drawdown numbers) and the Grid etc.

I wouldn’t call that incorrect.

More specifically, I don’t argue that intraday drawdowns are unimportant, but the program is explicitely designed from the beginning to deal with either daily or monthly data. I don’t think that it is valid to use intraday data in that case. Similarly, intramonth drawdowns will not be considered in an analysis of monthly values.

Well, intraday drawdowns are very real on a real money account and in some cases they might be significant. I think it’s a big mistake to ignore them.

When you say the program, what are you referring to?

I agree with Panu. Intraday drawdowns are essential. It is our real money that is being drawndown as the minutes pass and subscribers must be able to determine how much your stomach can take throughout every minute that a trade is open. If true drawdowns are not being represented, they should be corrected as soon as possible.

I am referring to the “advanced statistics” program. The max drawdown in the orange / grey table on the systems page is not always the same, so perhaps it includes intraday drawdowns; I don’t know.



I am not saying that they are not real. Where did you get that from? I am not Pal.



What I’m saying is that it is not some kind of typo in the code, but a structural feature of the program. I’m not convinced that it should change. I’m not convinced that it should remain the same either. That intraday drawdowns are real is true, but it is no argument. The EOD values are real too. What matters is the original definition (in case of the Calmar ratio) and/ or the industry standard. I’ll consider it closer.

I googled "calmar ratio" and "maximum drawdown" and all I have seen is that, as far as the authors are specific, these statistics are computed on daily or even monthly values. See



http://www.andreassteiner.net/performanceanalysis/?Risk_Measurement:Absolute_Risk:Maximum_Drawdown



http://www.autumngold.com/Performance/DescriptionDD.htm



http://www.raisepartner.com/?rep=lexique&ido=31&maximum_drawdown



http://alumnus.caltech.edu/~amir/mdd-risk.pdf



That the maximum drawdown depends on the interval that you consider is pointed out by several authors:



http://www.turtletrader.com/drawdown.pdf



http://www.gloriamundi.org/picsresources/easj.pdf



The classical standard for risk measurement is the 1996 Riskmetrics technical document of Morgan / Reuters, and they explicitely state in section 4.1 that all their metrics are based on the log of daily returns:



http://www.riskmetrics.com/pdf/td4e.pdf



Morningstar also computes all statistics on basis of daily or even monthly values:



http://datalab.morningstar.com/Midas/PDFs/Research_StarRating.pdf



Thus, as far as the advanced statistics module is concerned, I see no reason to include intraday values in any statistic, unless you provide ample evidence that it is the industry standard to do so.

I think including intraday values in the advanced statistics will open a can of worms b/c you have to deal with a bimodal distribution (overnight returns and intraday returns).



I’d expect for most systems the additional drawdown during the day is relatively small compared to end-of-day drawdowns and not of great concern. But for some systems such as RSMD, the difference appears to be large. In that case people should understand that risk measures based on end-of-day data underestimate the real risk (“real” as in getting a margin call), and apply the advanced risk statistics in a conservative way.



To allow members to make that judgment, it would be helpful if C2 could report both a drawdown based on end-of-day values, and on intraday values. If members see a large difference between the two, they know they have to interpret the advanced (risk) statistics cautiously.

"it would be helpful if C2 could report both a drawdown based on end-of-day values, and on intraday values."



That seems the best solution to me. Perhaps it is already done if the MDD of the basic statistics table is based on intraday values. In that case I still don’t understand the example that you gave in the first post of this thread…

Maybe Matthew could clarify, is the max draw down based on intraday or EOD values? It would be good to have that popup help next to max DD label which would tell how it is calculated.



Jules, thanks for you efforts in providing adv stats.