Knowledge needed about drawdown

 
Hi,

With strategy tester I am trying to understand, that with drawdown if there are no losses then why is drawdown not zero? Check it out for yourself:


Bars in test13254Ticks modelled12706147Modelling quality90.00%
Mismatched charts errors1




Initial deposit5000.00



Total net profit5395.96Gross profit5395.96Gross loss-0.00
Profit factor
Expected payoff39.97

Absolute drawdown405.00Maximal drawdown949.00 (10.79%)Relative drawdown10.93% (756.00)

Total trades135Short positions (won %)60 (100.00%)Long positions (won %)75 (100.00%)

Profit trades (% of total)135 (100.00%)Loss trades (% of total)0 (0.00%)
Largestprofit trade75.00loss trade-0.00
Averageprofit trade39.97loss trade-0.00
Maximumconsecutive wins (profit in money)135 (5395.96)consecutive losses (loss in money)0 (-0.00)
Maximalconsecutive profit (count of wins)5395.96 (135)consecutive loss (count of losses)-0.00 (0)
Averageconsecutive wins135consecutive losses0
 

Drawdown includes floating losses which transpire during the life of any given trade, not just the realized losses. In other words drawdown is an account equity calculation not an account balance calc.

However drawdown is pretty much a useless metric for assessing risk or risk versus reward, see http://www.futuresmag.com/Issues/2009/August2009/Pages/Minimizing-your-risk-of-ruin.aspx

 
1005phillip:

Drawdown includes floating losses which transpire during the life of any given trade, not just the realized losses. In other words drawdown is an account equity calculation not an account balance calc.

However drawdown is pretty much a useless metric for assessing risk or risk versus reward, see http://www.futuresmag.com/Issues/2009/August2009/Pages/Minimizing-your-risk-of-ruin.aspx


If it represents risk it is almost showing that it will be positive. Something like "Expected Payoff" or something? -No losses. This stumps me because I am used to thinking that I should look at the maximal draw down for the percent and assume that this is the percent of my account balance representative of losses on the initial balance. Again, this stumps me because when I see losses 0 then I really wanted to see drawdown percent as zero. Also, if I am researching this correctly and its to show the risk associated on the account with regard to the percent of change from the last change I guess this can only mean the risk and percent is going upward and no actually a draw down. Thanks!
 
Subgenius:
This stumps me because I am used to thinking that I should look at the maximal draw down for the percent and assume that this is the percent of my account balance representative of losses on the initial balance.

No... I'll borrow the explanation from the article "What the Numbers in the Expert Testing Report Mean" (https://www.mql5.com/en/articles/1486) with one minor correction - instead of balance we use equity to calculate DD (the article is old; at the time of it's writing the platform calculated DD using balance, but it has been changed to equity a long time ago):

  • Absolute drawdown is the difference between the initial deposit and the smalles value of equity within testing:

    AbsoluteDrawDown = InitialDeposit - MinimalEquity
  • Maximal drawdown is the highest difference between one of local upper extremums of the equity graph and the following lower extremums:

    MaximalDrawDown = Max of (Maximal Peak - next Minimal Peak)

    The basic stages of changing the maximal drawdown value within testing are given in the picture below. The total maximal drawdown value is in the thick arrows.

  • The maximal drawdown percentage shows the ratio between the maximal drawdown and the value of respective local upper extremum (of equity):
MaxDrawDown % = MaxDrawDown / its MaxPeak * 100%

And here is an explanation of relative DD (which was not calculated in the Report at the time of writing that article, so it's not mentioned in the article). I'll quote myself from another post (https://www.mql5.com/en/forum/123865):

Relative drawdown percentage (#1 in the figure) is the max value of maximal drawdown in percentage (#3 in the figure) achieved during the simulation. Relative drawdown (#2) is the corresponding drawdown achieved.

The most important of these figures IMHO is relative DD percentage (%).
 

We seem to be forgetting a simple factor here. It would be my assymption that the tester is smart enough to calculate this into the estimate of production. That is swap costs. When trades are open across market opens/closes, the swap is either a fee or a bonus. In a demo or live trade scenario, it is reflected a part of the drawdown when trade is still open and is further calculated in as part of the loss when trades are closed.

 
LEHayes:

We seem to be forgetting a simple factor here. It would be my assymption that the tester is smart enough to calculate this into the estimate of production. That is swap costs. When trades are open across market opens/closes, the swap is either a fee or a bonus. In a demo or live trade scenario, it is reflected a part of the drawdown when trade is still open and is further calculated in as part of the loss when trades are closed.


Hi. I will check that out. I have been trying to get historical data and I will post my results. Does anyone have or know where to get historical data as far back as possible? Thanks!
 
Subgenius:


Hi. I will check that out. I have been trying to get historical data and I will post my results. Does anyone have or know where to get historical data as far back as possible? Thanks!
SymbolEURUSD (Euro vs US Dollar)
Period30 Minutes (M30) 2001.01.01 07:00 - 2009.12.30 23:30 (2001.01.01 - 2009.12.31)
ModelEvery tick (the most precise method based on all available least timeframes)
ParametersFixedLotSize=0.10;

Bars in test112703Ticks modelled32343703Modelling qualityn/a
Mismatched charts errors155




Initial deposit5,000.00



Total net profit30,718.37Gross profit30,993.02Gross loss-274.66
Profit factor112.84Expected payoff38.64

Absolute drawdown79.00Maximal drawdown3328.27 (12.33%)Relative drawdown19.32% (2498.29)

Total trades795Short positions (won %)330 (100.00%)Long positions (won %)465 (98.28%)

Profit trades (% of total)787 (98.99%)Loss trades (% of total)8 (1.01%)
Largestprofit trade75.00loss trade-34.33
Averageprofit trade39.38loss trade-34.33
Maximumconsecutive wins (profit in money)465 (18368.31)consecutive losses (loss in money)8 (-274.66)
Maximalconsecutive profit (count of wins)18,368.31 (465)consecutive loss (count of losses)-274.66 (8)
Averageconsecutive wins394consecutive losses8

 

Given so few positions closed for a loss but the account experienced nearly 20% relative drawdown the strategy must be running with very wide stops (or none at all) and I would expect the MAE versus P/L analysis to be rather telling.

https://www.mql5.com/en/articles/1492

https://www.mql5.com/en/articles/mt4/c/articles/2007/08/MAE_Distribution.gif

 
1005phillip:

Given so few positions closed for a loss but the account experienced nearly 20% relative drawdown the strategy must be running with very wide stops (or none at all) and I would expect the MAE versus P/L analysis to be rather telling.

https://www.mql5.com/en/articles/1492

That is an excellent article. Thanks very much for highlighting it.

 
gordon:

No... I'll borrow the explanation from the article "What the Numbers in the Expert Testing Report Mean" (https://www.mql5.com/en/articles/1486) with one minor correction - instead of balance we use equity to calculate DD (the article is old; at the time of it's writing the platform calculated DD using balance, but it has been changed to equity a long time ago):


And here is an explanation of relative DD (which was not calculated in the Report at the time of writing that article, so it's not mentioned in the article). I'll quote myself from another post (https://www.mql5.com/en/forum/123865):
The most important of these figures IMHO is relative DD percentage (%).


Hello,

I would prefer some clearer definitions of the different "drawdowns" DDs (equity instead of balance is correct):

1. DDs in $ are all DDs in the graphic with number1, 2, 3 and 4 (the DD with thick arrow), that means DD peak Equity / next minimum Equity

2. the MAXIMUM DD in $ of all DDs is the DD nr. 4 = max(all DDs in $)

3. a RELATIVE DD in % is a DD / DD peak Equity * 100

4. the RELATIVE MAXIMUM DD is the maximum DD / maximum DD peak Equity * 100, this is the value in (...%)

now be careful:

5. in the report: the relative DD in % is the maximum of all relative DDs in % = max(all rel. DDs in %), this expression should better named as MAXIMUM RELATIVE DD in %

6. in the report: the relative DD in $ is the max(all rel. DDs in $) = MAXIMUM RELATIVE DD in % * maximum DD peak Equity / 100, this is the value in (...)

In the description of the report the word "relative" is used for different meanings!

 

Sorry, please correct line 6.

6. in the report: the relative DD in $ is the MAXIMUM RELATIVE DD in % * this DD peak Equity / 100, this is the value in (...)

The maximum relative DD % is NOT the same as the relative maximum DD %, also for $ instead of %!

Reason: