Most of us are familiar with different performance metrics, especially the ones in the MT4 reports.
Many of us look at DD Max $ and DD Relative % and measure it against the total net profit to give us a feel for Risk/Return.
Is there a metric that measures the relationship between Max DD in $ vs. Total net profit in $? And then compares it to the max DD in %?
If so, what is it called? And is it effective and a consistent measurement of risk adjusted return?
For example, comparing three EAs with 1000 currency units invested for simplicity's sake here are the results (the first two are very similar results of two quite different EAs I'm currently testing, the 3rd is a random EA).
EA1. Total net profit 400, max dd $ 100, max dd percent 10%
a. Max DD $ vs NP$ = 0.25 (100/400) lower is better
b. NP% vs Max DD% ratio = 4/1 (40%/10%) higher is better
EA2. Total net profit 40000, max dd $ 20000, max dd percent 40%
a. Max DD $ vs NP$ = 0.50 (20000/40000) lower is better
b. NP% vs DD% ratio = 100/1 (4000%/40%) higher is better
and a 3rd random riskier R/R martingale type EA that has yet to completely implode, but has stretched the DD to 80% and has so far yielded a good ROI...
EA3. Total net profit 4000 (10x better net than EA1), max dd $ 9000, max dd percent 80%
a. Max DD $ vs NP$ = 2.25 (9000/4000) lower is better
b. NP% vs DD% ratio = 5.0/1 (400%/80%) higher is better
By the first looks of comparing the different EA results, there are two ways to think about them...
EA 1 is conservative and has a nice low DD% of 10% and nice realistic return of 40%, has a solid winning trades vs losing trades ratio of over 50%.
EA 2 has a huge DD% and is high risk, but could be a nice gamble to throw a few dollars at, even though it shows a W/L ratio of less than 50%.
EA 3 shows a solid return and although DD is high, shows a W/L of over 80%, because it either trades with huge SLs or hedges losses and waits it out, or takes small profits quickly.
Many would choose A because the DD is lower and thus "safer", yet it yields a nice realistic annual return.
Others would "gamble" on B, even though it was unrealistic. Many would ignore B and think the greater risk/return is impossible or is too much of a gamble not worth risking 40% of capital for such an unrealistic return.
Others would choose C because it has a 400% and a W/L ratio of over 80%.
ALL assumptions are incorrect.
When the EAs become adjusted to trade the equivalent amount of riskDD% returned, the results becomes quite different.
EA has the lowest risk at first glance. However, EA 2 has twice the reward of EA1 if you adjust to the same amount of DD% risk. In other words, if EA2 was adjusted to trade the amount sizes to yield the same DD percentage of 10%, it would have yielded about a 800 unit return in the same period with about a 10% risk, instead of the 400 unit return for the same risk of EA1. EA3 has 10x the return of EA1, but if adjusted for the same amount of risk, actually has a lower return.
I know there's a Sharpe ratio, but I'm thinking there is an easier way to compare EA results assuming my maths are correct and assuming it works for all EA result comparisons. For now, I've come up with a formula for risk adjusted return that goes like this...
1 / ( (MaxDD$/NP$) / MaxDD% )
For EA1 this would be: 1 / ((100 / 400) / 0.10) = 1/(0.25/0.10) = 1/2.5 = 0.4
For EA2 this would be: 1 / ((20000 / 40000) / 0.40) = 1/(0.50/0.40) = 1/1.25 = 0.8
The 3rd random EA would be: 1 / ((9000 / 4000) / 0.80) = 1/(2.25/0.80) = 1/2.8125= 0.355
It shows that EA 2 is indeed twice as powerful as EA1 and as long as I was able to adjust the position sizes accordingly that my reward would be 2x more for the same amount of risk under the same trading scenarios. It also shows that EA3 returned 10x net profits compared to EA1 with less than 10x the overall risk of EA1, but that EA1 actually has a better return adjusted for risk compared to EA3.
Unless you have unlimited resources, at any point in time, you'll need to choose one EA or method over others to actually trade.
After all has been analyzed and compared, could this metric, if statistically sound, be another way to filter and find the better risk adjusted EA or method?
Comments? Critique? Errors?
Other ideas/implementations/derivatives of ways to better measure currently existing and known EA metrics are open for discussion.
Many of us look at DD Max $ and DD Relative % and measure it against the total net profit to give us a feel for Risk/Return.
Is there a metric that measures the relationship between Max DD in $ vs. Total net profit in $? And then compares it to the max DD in %?
If so, what is it called? And is it effective and a consistent measurement of risk adjusted return?
For example, comparing three EAs with 1000 currency units invested for simplicity's sake here are the results (the first two are very similar results of two quite different EAs I'm currently testing, the 3rd is a random EA).
EA1. Total net profit 400, max dd $ 100, max dd percent 10%
a. Max DD $ vs NP$ = 0.25 (100/400) lower is better
b. NP% vs Max DD% ratio = 4/1 (40%/10%) higher is better
EA2. Total net profit 40000, max dd $ 20000, max dd percent 40%
a. Max DD $ vs NP$ = 0.50 (20000/40000) lower is better
b. NP% vs DD% ratio = 100/1 (4000%/40%) higher is better
and a 3rd random riskier R/R martingale type EA that has yet to completely implode, but has stretched the DD to 80% and has so far yielded a good ROI...
EA3. Total net profit 4000 (10x better net than EA1), max dd $ 9000, max dd percent 80%
a. Max DD $ vs NP$ = 2.25 (9000/4000) lower is better
b. NP% vs DD% ratio = 5.0/1 (400%/80%) higher is better
By the first looks of comparing the different EA results, there are two ways to think about them...
EA 1 is conservative and has a nice low DD% of 10% and nice realistic return of 40%, has a solid winning trades vs losing trades ratio of over 50%.
EA 2 has a huge DD% and is high risk, but could be a nice gamble to throw a few dollars at, even though it shows a W/L ratio of less than 50%.
EA 3 shows a solid return and although DD is high, shows a W/L of over 80%, because it either trades with huge SLs or hedges losses and waits it out, or takes small profits quickly.
Many would choose A because the DD is lower and thus "safer", yet it yields a nice realistic annual return.
Others would "gamble" on B, even though it was unrealistic. Many would ignore B and think the greater risk/return is impossible or is too much of a gamble not worth risking 40% of capital for such an unrealistic return.
Others would choose C because it has a 400% and a W/L ratio of over 80%.
ALL assumptions are incorrect.
When the EAs become adjusted to trade the equivalent amount of riskDD% returned, the results becomes quite different.
EA has the lowest risk at first glance. However, EA 2 has twice the reward of EA1 if you adjust to the same amount of DD% risk. In other words, if EA2 was adjusted to trade the amount sizes to yield the same DD percentage of 10%, it would have yielded about a 800 unit return in the same period with about a 10% risk, instead of the 400 unit return for the same risk of EA1. EA3 has 10x the return of EA1, but if adjusted for the same amount of risk, actually has a lower return.
I know there's a Sharpe ratio, but I'm thinking there is an easier way to compare EA results assuming my maths are correct and assuming it works for all EA result comparisons. For now, I've come up with a formula for risk adjusted return that goes like this...
1 / ( (MaxDD$/NP$) / MaxDD% )
For EA1 this would be: 1 / ((100 / 400) / 0.10) = 1/(0.25/0.10) = 1/2.5 = 0.4
For EA2 this would be: 1 / ((20000 / 40000) / 0.40) = 1/(0.50/0.40) = 1/1.25 = 0.8
The 3rd random EA would be: 1 / ((9000 / 4000) / 0.80) = 1/(2.25/0.80) = 1/2.8125= 0.355
It shows that EA 2 is indeed twice as powerful as EA1 and as long as I was able to adjust the position sizes accordingly that my reward would be 2x more for the same amount of risk under the same trading scenarios. It also shows that EA3 returned 10x net profits compared to EA1 with less than 10x the overall risk of EA1, but that EA1 actually has a better return adjusted for risk compared to EA3.
Unless you have unlimited resources, at any point in time, you'll need to choose one EA or method over others to actually trade.
After all has been analyzed and compared, could this metric, if statistically sound, be another way to filter and find the better risk adjusted EA or method?
Comments? Critique? Errors?
Other ideas/implementations/derivatives of ways to better measure currently existing and known EA metrics are open for discussion.
