# Modified Dietz still a valid method to calculate composite return?

After 1/1/2011, is Modified Dietz still a valid method to calculate Composite Return? -Sorry, I post it again as a single question.

According to GIPS it doesn’t seem. No approximation are accepted anymore.

But it is still available for a question of prior returns. Don’t punt on it just because going forward it is not acceptable.

After 2010 , you have to do a valuation at each major cash flow event , so the need for Dietz doesn’t arise( i.e. you can use the TWRR instead of the MWRR)

Hi, guys. I know you’ve done GIPS long time ago… I’d like to know the validity of Modified Dietz method in COMPOSITE return calculation. Modified Dietz method is no longer valid for Portolio return calculation. Thanks.

COMPOSITE Return is an asset weighted average of the included portfolios ( which are valued at least monthly) . Do the portfolios using the right methods and then get the composite return as a weighted average

Thanks, janakisri. That’s exactly what I’d like to know. Basically, we can use Modified Dietz method to calculate the composite return by considering the composite (multiple portfolios in it) as one “super” portfolio. Schweser uses it on Notes Book 5, Page 194. CFAI Curriculum uses it on V6, page 296; and it seems valid.

Thanks deriv , I see your point now .

Don’t think it is right. Your method assumes modified Dietz which is no longer allowed. I believe that after 2010, you have the following options at composite level: Do true TWR (not modified Dietz) at least monthly for each portfolio. 1. Average using start value weighted 2. Average using start value +cash flow weighted. 3. Treat composite as a super portfolio and do a true TWR on it (not modified Dietz). deriv108 Wrote: ------------------------------------------------------- > Thanks, janakisri. That’s exactly what I’d like to > know. > > Basically, we can use Modified Dietz method to > calculate the composite return by considering the > composite (multiple portfolios in it) as one > “super” portfolio. > > Schweser uses it on Notes Book 5, Page 194. > CFAI Curriculum uses it on V6, page 296; and it > seems valid.

Hang on a sec, after 1/1/2010, for “portfolio valuation” you can still use modified Dietz if cashflows are NOT LARGE (CFAI text page 284-285) and firms must value portfolios monthly. For Large CF (no quantitative guideline given), porfolios must be valued as of the date of the CF (so no modified dietz for LARGE cashflows). Post 1/1/2010, there is RECOMMENDATION and not a REQUIREMENT that porfolios must be valued as of the date of all external CF. For Composite calculation, there are two methods (same as before) beg of period and beg of period + CF.

onelasttime Wrote: ------------------------------------------------------- > Hang on a sec, after 1/1/2010, for “portfolio > valuation” you can still use modified Dietz if > cashflows are NOT LARGE (CFAI text page 284-285) > and firms must value portfolios monthly. > > For Large CF (no quantitative guideline given), > porfolios must be valued as of the date of the CF > (so no modified dietz for LARGE cashflows). > Agree. Further on, the standard differentiates between large CF and significant CF which you must treat specially. No modified Dietz at all, since if it is not large, you just > Post 1/1/2010, there is RECOMMENDATION and not a > REQUIREMENT that porfolios must be valued as of > the date of all external CF. > Not sure what you mean, but from 2010 “firms must calculate performance for interim periods between all large external cash flows and geometrically link performance to calculate period returns.” i.e., true TWR a must for all large CF. Again referring to Large external CF comment earlier. > For Composite calculation, there are two methods > (same as before) beg of period and beg of period + > CF. No, actually there are three methods as I mentioned. The super method is mentioned as the “aggregate return method” http://www.gipsstandards.org/standards/guidance/archive/pdf/GSCalcMethRevised.pdf

> > Post 1/1/2010, there is RECOMMENDATION and not > a > > REQUIREMENT that porfolios must be valued as of > > the date of all external CF. > > > Not sure what you mean, but from 2010 “firms must > calculate performance for interim periods between > all large external cash flows and geometrically > link performance to calculate period returns.” > i.e., true TWR a must for all large CF. Again > referring to Large external CF comment earlier. > Page 285 (middle of the page) Post 1/1/2010, there is a “recommendation” to value portfolios on ALL external cashflows not just large. Agree with the three methods on composite construction, the third method is really just a variation of the 2nd method (mathematically). Page 296 CFAI text. The results under method 2 and 3 are the same.

elcfa Wrote: ------------------------------------------------------- > Not sure what you mean, but from 2010 “firms must > calculate performance for interim periods between > all large external cash flows and geometrically > link performance to calculate period returns.” > i.e., true TWR a must for all large CF. Again > referring to Large external CF comment earlier. The question point shall be : how to WEIGHT the return of each portfolio in a composite from 2010/1/1 ? For portfolio return, it shall be correct that TRUE TWR shall be used. Do you mean the cash flows of all portfolios shall be treated as in a composite (a portfolio of portfolios) ? For example, assume all interim cash flows are “large” and Portfolio A April 1 : initial value, 500M. April 10 : cash inflow, 60M. April 20 : cash outflow, 10M Portfolio B April 1 : initial value, 300M. April 10 : cash inflow, 30M. Then treat portfolio A & B as a composite (aggregated) as follows ? April 1 : initial value, 800M. April 10 : cash inflow, 90M. April 20 : cash outflow, 10M

stop confusing everyone

SkipE99 Wrote: ------------------------------------------------------- > stop confusing everyone We are trying to clarify this issue to avoid confusion rather than increasing confusion. Please just skip if you are not concerned about this issue.

elcfa, I am sorry. Would you please kindly advise if TRUE TWR shall be calculated (rather than return calculated Modified Dietz method) in the composite example raised in my previous message. I think quite a lot of candidates are confused by this issue. Thanks in advance !

alta168 My read on the new GIPS give me understanding that you must use TRUE TWR after 2011, not modified Dietz (as before 2011) on the composite. I must agree that GIPS is very vague in this aspect and one still can find old examples on the GIPS site showing aggregate return using modified Dietz. Here are a couple of quotes from the new GIPS. “For periods beginning 1 January 2010, at the latest, firms must calculate performance for interim periods between all large external cash flows and geometrically link performance to calculate period returns” for portfolio returns. “Composite returns must be calculated by asset weighting the individual portfolio returns using beginning-of-period values or a method that reflects both beginning-of-period values and external cash flows” “The Aggregate Return method combines all the composite assets and external cash flows before any calculations occur to calculate returns as if the composite were ONE PORTFOLIO. The method is also acceptable as an asset-weighted approach” If GIPS were still to allow usage of (sneak) modified Dietz on composite while insisting true TWR on portfolio, it would beat the purpose of introducing true TWR because of three reasons: 1. One only sees composite return in the GIPS report, not portfolio return, so why be stricter on something you don’t see while accepting a more lax standard at something everyone sees? 2. Gross inconsistency between composite vs. portfolio level. 3. Misleading wording in the GIPS standard: “as if the composite were ONE PORTFOLIO” give the expectation that the portfolio level method applies. My two cents. I may be wrong. onelasttime wrote >Agree with the three methods on composite construction, the third method is really just > a variation of the 2nd method (mathematically). Page 296 CFAI text. The results under >method 2 and 3 are the same. Method 2 and 3 are the same if you are applying the modified Dietz calculation at portfolio level (i.e., before 2011), but they would yield different results if you use true TWR.

elcfa, Thanks a lot for your further advice. So, shall we conclude that “true TWR” shall be applied to composite (as in the example raised by me) ? Thank you so much ! On the other hand, some candidates (include me) have a question concerning if MIRR = MWRR and if TWR = true TWR ? Below is the message posted by me. Would you please give us your advice too ? You may refer to original post too. Re: What if the difference between MIRR and MWRR Posted by: alta168 (IP Logged) Date: April 29, 2011 06:58AM Please Refer to CFAI text Vol 6 In Example 4 (P.130) & Example 5, the returns calculated by TWR & MWR are almost same. Furthermore, the footnote 19 at the bottom of P.286 stated that “MIRR” approximates a TWR. But the return (0.41%) calculated using True TWR and the return (0.51%) calculated using MIRR (MWRR?) are quite different. (Remark : Please refer to P.285 too) Now there are 2 questions : 1. MIRR = MWRR ? 2. True TWR = TWR ? Why it is said “True” ?

elcfa, Very sorry, the message posted by me shall be following one. Re: What if the difference between MIRR and MWRR Posted by: AMA (IP Logged) Date: April 28, 2011 09:06PM I think these two shall be same. Maybe you can verify by calculating a longer period’s return.

AMA There are mainly two weighting methods/categories: time weighting or money weighting. In each of those categories, there are many techniques. Time weighted (TWR): True TWR, original Dietz, Modified Dietz. The last two are some sort of quick and dirty TWR calculation since true TWR is quite demanding (you need to find the market value for each asset in the portfolio every time you have a large CF) and you need to do a lot more computation. Money weighted (MWR): You have essentially various IRR (internal rate of return) based methods under this category (standard IRR, modified IRR, Linked IRR (also known as the BAI method,…) Because of its modifications, the modified IRR, Linked IRR approximate the TWR returns. Hope it is clearer.