Acceptable Drawdown

 

Hi Pors.

How large a maximum drawdown(in percent) is generally acceptable to you? Do you have a guideline?

 

The concept of drawdown as it is applied as a means of assessing risk is inherently flawed:

Few traders know how to calculate risk of ruin. Many traders talk about maximum drawdown, as if there is a maximum limit. Obviously, once the “maximum” drawdown has happened, further losses are still possible. The maximum drawdown is merely the point at which the bad luck ended in some historical data set. The maximum observed drawdown will continue to increase the longer the game is played.

Source: Minimizing your risk of ruin

You really want to pursue established VAR methods for systematic risk assessment and management. Risk of ruin calcs help you capture and express your expected drawdown assuming stationarity of the trade strategy's rate of return and variance.

 
1005phillip:

The concept of drawdown as it is applied as a means of assessing risk is inherently flawed:

You really want to pursue established VAR methods for systematic risk assessment and management. Risk of ruin calcs help you capture and express your expected drawdown assuming stationarity of the trade strategy's rate of return and variance.

But the drawdown is defaultly given out by the meta4 backtest. it is convenient to use, and it is a statistic more or less providing some information......right?
 

Drawdown is a metric given by the strategy tester but that doesn't make it useful or relevant.

For example look at the max consecutive wins and max consecutive losses metrics...how relevant is that info? Are you to assume that in the future you are going to experience a successful win-rate or loss-rate comparable to that produced on historical data?

That would only happen if the future market behaves effectively identical to the past market. When has that ever been the case?

The question you need to ask yourself is what purpose are you aiming to satisfy by doing backtesting? If the answer is "I want to use past results created by the backtester as an indication of potential future results" then you have your work cut out for you.

I do backtesting as a means of debugging my code as well as a means of eliminating truly awful strategies and parameter values.

 
1005phillip:

Drawdown is a metric given by the strategy tester but that doesn't make it useful or relevant.

For example look at the max consecutive wins and max consecutive losses metrics...how relevant is that info? Are you to assume that in the future you are going to experience a successful win-rate or loss-rate comparable to that produced on historical data?

That would only happen if the future market behaves effectively identical to the past market. When has that ever been the case?

The question you need to ask yourself is what purpose are you aiming to satisfy by doing backtesting? If the answer is "I want to use past results created by the backtester as an indication of potential future results" then you have your work cut out for you.

I do backtesting as a means of debugging my code as well as a means of eliminating truly awful strategies and parameter values.


If your system in backtesting is very profitable, has very small variance. but the max drawdown is above 50%, are you confindent to use it ?

If you say such a excellent system could not produce a drawdown as that high, then that means the drawdown is still correlated to the variance..........

 
Not to confuse you, ther's relative and absolute drawdown. I'm never comfortable with drawdown more than 20%, especially together with huge consecutive losses.
 
IMO Drawdown is a relative term. All of your statistical measurements are relative. Example, if you choose 20% as your acceptable Max/Rel Drawdown. Would you rule out a strong performing system because the Drawdown is 21%. Also, the Drawdown would be different depending on when you decided to stop the simulator i.e depending on what time period you test for.  Any Statistical data can be skewed. Forex is a game of Options not Decisions. i.e Broker A gives you better options than Broker B. Your goal is to develop Systems which gives you the best options, i.e System A gives me less Drawdown than System B. When you reach your limit and cannot out perform the previous system anymore. Your only other option is Diving into the market with NO guarantees.
 
zbleach:


If your system in backtesting is very profitable, has very small variance. but the max drawdown is above 50%, are you confindent to use it ?

If you say such a excellent system could not produce a drawdown as that high, then that means the drawdown is still correlated to the variance..........


Depends on what caused the drawdown. Depending on how you have engineered your trade strategy, drawdown is simply the intersection of coincidence and market timing over a historical time period.

Unless you expect future time periods to replicate these coincidences and have engineered your strategy to embark on similar market timing then future drawdown will have little to do with historical drawdown.

I don't measure drawdown in backtesting. I look for plateau's of stationarity of the market opportunities (coincidences) and my strategy's ability to take advantage of them (market timing).

I do measure the historical variance produced from these plateaus for risk of ruin assessments but that is about it.