GIPS compliance (yes / no) 1. this composite was created in March 2004, but reproting historty starts from 2003. (yes) 2. all accounts of at least $10,000,000 are included in composite as of the first quarter under management. (yes?) 3. dispersion is measured by annualized monthly standard deviation of portfolio returns for portfolios that were in the composite over the entire year. (yes?) 4. presented amount of assets and number of portfolios in the composite, but without total firm assets and the percentage of firm assets represented by the composite. (no?) 5. composite returns are geometirically linked quarterly returns calculated by asset-weighting the individual portfolio returns computed with modified Dietz method. (yes) 6. performance results are presented gross-of-fee after deduction of trading expenses computed with implementation shortfall method. relavant fee schedule is attached. (no) I gave out my vote. like to see yours.
- no - can’t backdate 2. no - fee paying discretionary is what matters, not portfolio size 3. no - dispersion of accounts in the composite (time frame doesn’t matter?) 4. no - need total firm assets and the percentage of the composite to total firm assets 5. ? i never can remember this one 6. yes. i think this is correct.
5- M. Deitz is still good, but Deitz is not, Geometrically link monthly to quarterly, quarterly to yearsly.
strikershank Wrote: ------------------------------------------------------- > 1. no - can’t backdate GIPS asked for since inception or 5 year > 2. no - fee paying discretionary is what matters, > not portfolio size but, is an artificial threshold against the rule. if not, then. it’s a compliance. > 3. no - dispersion of accounts in the composite > (time frame doesn’t matter?) so, it’s a compliance then? > 4. no - need total firm assets and the percentage > of the composite to total firm assets > 5. ? i never can remember this one > 6. yes. i think this is correct. but, implementation shortfall gives you more than explict trading cost.
I assumed you had the answer - i’m going from memory as well. If you don’t have the answers then we should look up the correct versions before this thread grows and gets filled with wrong information.
no. i dont have answers. those are cases that i summrized myself. of course, i put some thoughts in it, but, i like conformations from audience.
ok - well the reason i picked what i did in greater detail. 1. if the composite was created in 2004, i don’t think you can claim compliance since 2003. GIPS wants 5 years, but i don’t think you can backfill information for composites. This differs from say a composite that was created in 2002 and then you claim compliance from 2003 onwards (or 5 years worth). 2. All discretionary fee paying accounts must be included. I believe this is hard and fast. NOW a firm might have a rule where it only takes fee paying discretionary accounts of $10mln or more and then change it to $2mln but when it adjusts the composite to include the new $2mln threshold i think it has to say the composite has changed. BUT what remains is that all fee paying discretionary must be included. 3. I don’t know. just doesn’t look right but need to verify. 4. need total firm assets and the percentage of the composite to total firm assets 5. ? 6. you make a good point - implementation shortfall shouldn’t be included…so i think i have changed my mind to a no on this one.
strikershank Wrote: ------------------------------------------------------- > ok - well the reason i picked what i did in > greater detail. > > 1. if the composite was created in 2004, i don’t > think you can claim compliance since 2003. GIPS > wants 5 years, but i don’t think you can backfill > information for composites. This differs from say > a composite that was created in 2002 and then you > claim compliance from 2003 onwards (or 5 years > worth). > i am with you. but, i just didn’t see GIPS is explicit on this. someone could run his own money for a year before lauching it. as long as he discloses the fact, would gips still asks him to take that year out? > 2. All discretionary fee paying accounts must be > included. I believe this is hard and fast. NOW a > firm might have a rule where it only takes fee > paying discretionary accounts of $10mln or more > and then change it to $2mln but when it adjusts > the composite to include the new $2mln threshold i > think it has to say the composite has changed. BUT > what remains is that all fee paying discretionary > must be included. > yes, i think your case is clear. but, one schweser practice question in book 6 says it’s fine. no violation. > 3. I don’t know. just doesn’t look right but need > to verify. the GIPs asks for standard deviation of annual returns. it most likely will not be the same as annualized std of monthly returns. but, for composite with very short history (say 5 years, 5 observations), annualizing (60 observations) yields more reliable results. > 5. ? just not sure what should be right wordings for return calculation in the footnotes. according to schweser, saying “return calculation is performed as prescribed by the GIPs” is not acceptable.
- This part really confuses me. In the input part there are monthly returns required. In the composites part quarterly weighted assets are required. So I am concluding that returns are monthly but asset weighting is quarterly. I don’t know whether I am making any sense.
Hiya Turkiya Wrote: ------------------------------------------------------- > 5. This part really confuses me. In the input part > there are monthly returns required. In the > composites part quarterly weighted assets are > required. So I am concluding that returns are > monthly but asset weighting is quarterly. I don’t > know whether I am making any sense. we are here talking about the bear minimum to pass GIPS rulings. it shouldn’t stop someone conducting composite calculations monthly. is that what you asking?
Not exactly, I am talking about bear minimum to pass CFA LIII:) To calculate composite return, there are two ways You can calculate one by one the return on portfolios in the composite and then take a weighted awerage of them to find the composite return. Returns on portfolios should be calculated monthly, but portfolios in the composites can be weighted quarterly. The other way is to treat the composite as if it is one portfolio. This way I have no issue. But first one I find confusing.
i dont see inconsistency in GIPS though. it requires monthly valuation but quarterly total return calculation. which part relates to the confusion you have?
rand0m Wrote: ------------------------------------------------------- > strikershank Wrote: > > 2. All discretionary fee paying accounts must > be > > included. I believe this is hard and fast. NOW > a > > firm might have a rule where it only takes fee > > paying discretionary accounts of $10mln or more > > and then change it to $2mln but when it adjusts > > the composite to include the new $2mln threshold > i > > think it has to say the composite has changed. > BUT > > what remains is that all fee paying > discretionary > > must be included. > > > > yes, i think your case is clear. but, one schweser > practice question in book 6 says it’s fine. no > violation. I agree with Schweser that there is NO violation. See 4.A.3. Even though this minimum asset level is exceptionally large, there is no violation here. (yes, any change to this minimum asset level must be disclosed as well) - sticky
rand0m Wrote: > the GIPs asks for standard deviation of annual > returns. agree. > it most likely will not be the same as > annualized std of monthly returns. but, for > composite with very short history (say 5 years, 5 > observations), annualizing (60 observations) > yields more reliable results. I don’t think projected performance like this is allowed in GIPS. No historical data should be used to come up to this year’s return or dispersion. - sticky
rand0m Wrote: ------------------------------------------------------- > 5. composite returns are geometirically linked > quarterly returns calculated by asset-weighting > the individual portfolio returns computed with > modified Dietz method. (yes) I think it’s a NO if I understand this correctly. In the first hand, a composite return should be calculated by asset-weighting the (annual) portfolio returns (2.A.3) and the “geometrical linking” is wrong as the first step. Comments? - sticky
Hiya Turkiya Wrote: ------------------------------------------------------- > To calculate composite return, there are two ways > > You can calculate one by one the return on > portfolios in the composite and then take a > weighted awerage of them to find the composite > return. agree. > Returns on portfolios should be > calculated monthly, but portfolios in the > composites can be weighted quarterly. What does that mean? I think according to 2.A.3, you MUST use beginning-of-period weighting of the portfolios to come up to the composite return. > The other way is to treat the composite as if it > is one portfolio. This way I have no issue. No. I don’t think you can do so. How many portfolios will you report under this composite then? BTW, I don’t think GIPS allows combining portfolios as a single new portfolio under a composite. As a side info, it only allows carve-outs. - sticky
sticky Wrote: ------------------------------------------------------- > rand0m Wrote: > -------------------------------------------------- > ----- > > > 5. composite returns are geometirically linked > > quarterly returns calculated by asset-weighting > > the individual portfolio returns computed with > > modified Dietz method. (yes) > > I think it’s a NO if I understand this correctly. > In the first hand, a composite return should be > calculated by asset-weighting the (annual) > portfolio returns (2.A.3) and the “geometrical > linking” is wrong as the first step. > > Comments? > > - sticky the total return needs to be calculated at least quarterly. to get annul return, of course, you have to link them. otherwise, how? I dont think you can get a better accuracy than that.
sticky Wrote: ------------------------------------------------------- > rand0m Wrote: > > > the GIPs asks for standard deviation of annual > > returns. > > agree. > > > it most likely will not be the same as > > annualized std of monthly returns. but, for > > composite with very short history (say 5 years, > 5 > > observations), annualizing (60 observations) > > yields more reliable results. > > I don’t think projected performance like this is > allowed in GIPS. No historical data should be > used to come up to this year’s return or > dispersion. > > - sticky there’s no projection there. these are real standard deviaion. you seem to be way out of base.