I was reading the following paper from Adam Butler:
http://www.bpgassociates.com/docs/AdaptiveAssetAllocationAPrimer.pdf I saw the example on page 26, namely one with an 8% annualized volatility target. I was wondering if there's a goto package among the R/Finance community to solve problems of the type: max w'r s.t. w'Sw = target 0 <= w_i <= 1 for all i Sum(i=1...n)w_i = 1 I know that quadprog solves problems of the form of min w'r + lambda*w'Sw s.t. 0 <= w_i <= 1 for all i Sum(i=1...n)w_i = 1 I was wondering if there was a goto package for SOCP so that I could solve the first type of problem without needing to call a global optimizer. Thanks a lot. Ilya [[alternative HTML version deleted]] _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go. 
Ilya, Bernhard's cccp package should be the definitive socp package in R
now. cran.rproject.org/package=cccp Alexios On 08/07/2015 18:37, Ilya Kipnis wrote: > I was reading the following paper from Adam Butler: > > http://www.bpgassociates.com/docs/AdaptiveAssetAllocationAPrimer.pdf > > I saw the example on page 26, namely one with an 8% annualized volatility target. > > I was wondering if there's a goto package among the R/Finance community to solve problems of the type: > > max w'r > s.t. > w'Sw = target > 0 <= w_i <= 1 for all i > Sum(i=1...n)w_i = 1 > > I know that quadprog solves problems of the form of > min w'r + lambda*w'Sw > s.t. > 0 <= w_i <= 1 for all i > Sum(i=1...n)w_i = 1 > > I was wondering if there was a goto package for SOCP so that I could solve the first type of problem without needing to call a global optimizer. > > Thanks a lot. > > Ilya > > [[alternative HTML version deleted]] > > _______________________________________________ > [hidden email] mailing list > https://stat.ethz.ch/mailman/listinfo/rsigfinance >  Subscriberposting only. If you want to post, subscribe first. >  Also note that this is not the rhelp list where general R questions should go. > > _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go. 
Dear Ilya & Alexios,
given that the Alexios played the ball into my court, please find below a toy example to your problem: (the parma package is loaded for the data set) library(parma) library(cccp) data("etfdata") R = etfdata/lag(etfdata)1 R = na.omit(R * 100) N < ncol(R) S = cov(coredata(R)) Se < eigen(S) Sr < Se$vectors %*% diag(sqrt(Se$values)) %*% t(Se$vectors) all.equal(Sr %*% Sr, S, check.attributes = FALSE) mu < colMeans(R) q < mu ## Formulating constraints ## Nonnegativity & SOC nvec < rep(0, N) nno1 < nnoc(G = diag(N), h = nvec) soc1 < socc(F = Sr, g = nvec, d = nvec, f = 1.5) ## Budget constraint A < matrix(rep(1, N), nrow = 1, ncol = N) b < matrix(1, nrow = 1, ncol = 1) ## Solving problem Popt < cccp(q = q, A = A, b = b, cList = list(nno1, soc1)) w < getx(Popt) ## checking all(w > 0) all.equal(sum(w), 1.0) (retmax < getstate(Popt)["pobj"]) (sdrisk < drop(sqrt(crossprod(w, S) %*% w))) Best wishes, Bernhard Ursprüngliche Nachricht Von: RSIGFinance [mailto:[hidden email]] Im Auftrag von alexios Gesendet: Donnerstag, 9. Juli 2015 15:17 An: Ilya Kipnis; [hidden email] Betreff: Re: [RSIGFinance] What's the best package for SOCP in R? Ilya, Bernhard's cccp package should be the definitive socp package in R now. cran.rproject.org/package=cccp Alexios On 08/07/2015 18:37, Ilya Kipnis wrote: > I was reading the following paper from Adam Butler: > > http://www.bpgassociates.com/docs/AdaptiveAssetAllocationAPrimer.p > df > > I saw the example on page 26, namely one with an 8% annualized volatility target. > > I was wondering if there's a goto package among the R/Finance community to solve problems of the type: > > max w'r > s.t. > w'Sw = target > 0 <= w_i <= 1 for all i > Sum(i=1...n)w_i = 1 > > I know that quadprog solves problems of the form of min w'r + > lambda*w'Sw s.t. > 0 <= w_i <= 1 for all i > Sum(i=1...n)w_i = 1 > > I was wondering if there was a goto package for SOCP so that I could solve the first type of problem without needing to call a global optimizer. > > Thanks a lot. > > Ilya > > [[alternative HTML version deleted]] > > _______________________________________________ > [hidden email] mailing list > https://stat.ethz.ch/mailman/listinfo/rsigfinance >  Subscriberposting only. If you want to post, subscribe first. >  Also note that this is not the rhelp list where general R questions should go. > > _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go. ***************************************************************** Confidentiality Note: The information contained in this ...{{dropped:10}} _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go. 
On Thu, Jul 9, 2015 at 9:03 AM, Pfaff, Bernhard Dr.
<[hidden email]> wrote: > Dear Ilya & Alexios, > > given that the Alexios played the ball into my court, please find below a toy example to your problem: > (the parma package is loaded for the data set) > > library(parma) > library(cccp) Sadly, library(parma) works its way down to Rglpk which doesn't build coming from CRAN. Is there a better place to get that? * installing *source* package ‘Rglpk’ ... ** package ‘Rglpk’ successfully unpacked and MD5 sums checked ** libs /bin/sh: line 0: cd: GLPK: No such file or directory Makevars:10: recipe for target 'GLPK.ts' failed make: *** [GLPK.ts] Error 1 ERROR: compilation failed for package ‘Rglpk’ Thanks, Mark _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go.
markknecht@gmail.com

On Fri, Jul 10, 2015 at 6:58 AM, Mark Knecht <[hidden email]> wrote:
> On Thu, Jul 9, 2015 at 9:03 AM, Pfaff, Bernhard Dr. > <[hidden email]> wrote: >> Dear Ilya & Alexios, >> >> given that the Alexios played the ball into my court, please find below a toy example to your problem: >> (the parma package is loaded for the data set) >> >> library(parma) >> library(cccp) > > > Sadly, library(parma) works its way down to Rglpk which doesn't build > coming from CRAN. Is there a better place to get that? > > * installing *source* package ‘Rglpk’ ... > ** package ‘Rglpk’ successfully unpacked and MD5 sums checked > ** libs > /bin/sh: line 0: cd: GLPK: No such file or directory > Makevars:10: recipe for target 'GLPK.ts' failed > make: *** [GLPK.ts] Error 1 > ERROR: compilation failed for package ‘Rglpk’ > > Thanks, > Mark Rglpk requires that I install glpk in Linux first. With that done the example runs. Thanks, Mark _______________________________________________ [hidden email] mailing list https://stat.ethz.ch/mailman/listinfo/rsigfinance  Subscriberposting only. If you want to post, subscribe first.  Also note that this is not the rhelp list where general R questions should go.
markknecht@gmail.com

Free forum by Nabble  Edit this page 