Hi,
I'm looking at the results returned by the function DrawdownPeak and struggle to understand the calculation.
In the following example, I recalculate the return from drawdown and for the third period I found a difference (I'm aware the formula I'm using is only local for the example).
I expect using "geometric" rule for aggregating returns but cannot match the calculations
In this example the peak is 10% and the returns for the next 2 period are -2% and -1% which should lead to a cumulative return of (1.1 / 1.078 / 1.06722)
Q1. The return for the 3rd period from Peak should then be : 1.06722 / 1.10 -1 = 2.98% and not 2.9998%. Is the result I should expect? (it looks like the it does not compound when "geometric=TRUE")
Q2. I noticed that "geometric=TRUE" option does not make any difference. Is this expected?
Q3. Finally I was not able to standardized the decimal with options(digits=4). Is this normal?
Code for replication
library("PerformanceAnalytics")
ret_num<-c(0.10,-0.020,-0.010) # simulated returns - the peak is the first observations
ret_test<-xts(ret_num, Sys.Date()-3:1) # return in xts class
ddtp_test<-DrawdownPeak(ret_test,geometric=TRUE) # results for return for drawdowns to validate
cret_num<-cumprod(1+ret_num)-1 # validation by cumulative returns
ddtp_vet<-(1+cret_num)/(1+cret_num[1])-1 # replication of cumulative returns for drawdowns
cbind(ddtp_test,ddtp_vet) # to compare results - differences
The third row shows a difference - look small bu
ddtp_test ddtp_vet
[1,] 0.000000 0.0000
[2,] -0.020000 -0.0200
[3,] -0.029998 -0.0298
This looks small (not related to rounding) but in other cases the difference was more important and lead to difference in the drawdown period.
I reproduced the same values with version 3.1 and version 3.1.3
Package PerformanceAnalytics (1.4.3579) loaded.
version.string R version 3.1.3 (2015-03-09)
Thank you
Best Regards,
Charles
validation of DrawdownPeak in PerformanceAnalytics package
3 messages · Charles Duranceau, Peter Carl
Charles, DrawdownPeak probably shouldn't be in the namespace - thanks for pointing it out. You might be looking for PerformanceAnalytics:::Drawdowns, which probably *should* be in the namespace. Let me know if that works for you. pcc
On 03/23/2015 06:05 PM, Charles Duranceau wrote:
Hi,
I'm looking at the results returned by the function DrawdownPeak and struggle to understand the calculation.
In the following example, I recalculate the return from drawdown and for the third period I found a difference (I'm aware the formula I'm using is only local for the example).
I expect using "geometric" rule for aggregating returns but cannot match the calculations
In this example the peak is 10% and the returns for the next 2 period are -2% and -1% which should lead to a cumulative return of (1.1 / 1.078 / 1.06722)
Q1. The return for the 3rd period from Peak should then be : 1.06722 / 1.10 -1 = 2.98% and not 2.9998%. Is the result I should expect? (it looks like the it does not compound when "geometric=TRUE")
Q2. I noticed that "geometric=TRUE" option does not make any difference. Is this expected?
Q3. Finally I was not able to standardized the decimal with options(digits=4). Is this normal?
Code for replication
library("PerformanceAnalytics")
ret_num<-c(0.10,-0.020,-0.010) # simulated returns - the peak is the first observations
ret_test<-xts(ret_num, Sys.Date()-3:1) # return in xts class
ddtp_test<-DrawdownPeak(ret_test,geometric=TRUE) # results for return for drawdowns to validate
cret_num<-cumprod(1+ret_num)-1 # validation by cumulative returns
ddtp_vet<-(1+cret_num)/(1+cret_num[1])-1 # replication of cumulative returns for drawdowns
cbind(ddtp_test,ddtp_vet) # to compare results - differences
The third row shows a difference - look small bu
ddtp_test ddtp_vet
[1,] 0.000000 0.0000
[2,] -0.020000 -0.0200
[3,] -0.029998 -0.0298
This looks small (not related to rounding) but in other cases the difference was more important and lead to difference in the drawdown period.
I reproduced the same values with version 3.1 and version 3.1.3
Package PerformanceAnalytics (1.4.3579) loaded.
version.string R version 3.1.3 (2015-03-09)
Thank you
Best Regards,
Charles
[[alternative HTML version deleted]]
_______________________________________________ R-SIG-Finance at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.
Peter Carl 312.307.6346
1 day later
Hi Peter, Yes that works. Thank you I found also easier working directly with "table.Drawdowns" but wanted to understand the building blocks. Cheers -----Original Message----- From: Peter Carl [mailto:peter at braverock.com] Sent: Monday, March 23, 2015 6:48 PM To: Charles Duranceau; r-sig-finance at r-project.org Subject: Re: [R-SIG-Finance] validation of DrawdownPeak in PerformanceAnalytics package Charles, DrawdownPeak probably shouldn't be in the namespace - thanks for pointing it out. You might be looking for PerformanceAnalytics:::Drawdowns, which probably *should* be in the namespace. Let me know if that works for you. pcc
On 03/23/2015 06:05 PM, Charles Duranceau wrote:
Hi,
I'm looking at the results returned by the function DrawdownPeak and struggle to understand the calculation.
In the following example, I recalculate the return from drawdown and for the third period I found a difference (I'm aware the formula I'm using is only local for the example).
I expect using "geometric" rule for aggregating returns but cannot
match the calculations
In this example the peak is 10% and the returns for the next 2 period
are -2% and -1% which should lead to a cumulative return of (1.1 /
1.078 / 1.06722)
Q1. The return for the 3rd period from Peak should then be : 1.06722 /
1.10 -1 = 2.98% and not 2.9998%. Is the result I should expect? (it looks like the it does not compound when "geometric=TRUE") Q2. I noticed that "geometric=TRUE" option does not make any difference. Is this expected?
Q3. Finally I was not able to standardized the decimal with options(digits=4). Is this normal?
Code for replication
library("PerformanceAnalytics")
ret_num<-c(0.10,-0.020,-0.010) # simulated returns - the peak is the first observations
ret_test<-xts(ret_num, Sys.Date()-3:1) # return in xts class
ddtp_test<-DrawdownPeak(ret_test,geometric=TRUE) # results for return for drawdowns to validate
cret_num<-cumprod(1+ret_num)-1 # validation by cumulative returns
ddtp_vet<-(1+cret_num)/(1+cret_num[1])-1 # replication of cumulative returns for drawdowns
cbind(ddtp_test,ddtp_vet) # to compare results - differences
The third row shows a difference - look small bu
ddtp_test ddtp_vet
[1,] 0.000000 0.0000
[2,] -0.020000 -0.0200
[3,] -0.029998 -0.0298
This looks small (not related to rounding) but in other cases the difference was more important and lead to difference in the drawdown period.
I reproduced the same values with version 3.1 and version 3.1.3
Package PerformanceAnalytics (1.4.3579) loaded.
version.string R version 3.1.3 (2015-03-09)
Thank you
Best Regards,
Charles
[[alternative HTML version deleted]]
_______________________________________________ R-SIG-Finance at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.
-- Peter Carl 312.307.6346