# My greatest GIPS confusion

GIPS always sounds vague to me and it’s time I drilled down to its details. Please, your input is essential to me. Please also include GIPS reference number in your response. (Q1) 2A6 says you must calc composite returns at least quarterly. Is there corresponding rule for portfolio returns? (Note that 1A3 is about portfolio valuation, NOT portfolio return) (Q2) It’s now 2006 and I’ve got portfolio monthly returns for portfolios p1 to p4 of Composite A. That is, if return of p1 on month2 is denoted by r12, then I’ve got 12 returns: - p1: p11, p12, p13 - p2: p21, …, p23 - p3: … - p4: p41, …, p43 2A6 says I must have composite returns quarterly. So I use beginning-quarter values of p1 to p4 (say, v1 to v4) as weighting. Fine. Question is: what portfolio returns should I use to arrive at the quarterly return of composite A? Quarterly returns? Monthly returns? Others? What is the GIPS refererence for this? (Q3) At the end of year 2006, I need to show my annual performance of composite A (that table thing). With 4 quarterly composite returns already calculated for composite A over the year, how do I arrive at the annual return of composite A to be shown in the table? What is the GIPS reference for this? (Q4) (continued from Q3) For composite A, I’ve got 4 quarterly composite returns, I’ve got 12 monthly returns of p1-p4 (total 48 returns). How do I arrive at the internal dispersion of composite A to be shown in the table, as required by 5A1d? What is the GIPS reference for this? (Q5) (continued from Q3) I am only showing annual return of composite A in my presentation (the table). Why do I need to calculate quarterly return of composite A? It’s not shown anyhow. - sticky

Time Value return calculate the monthly returns, geometrically connect them to form quarterly, do the same with the quarterly returns to form the annual ones… DONE… oh and don’t annualize partial returns. Internal Dispersion, can be high/low, STDev, etc…

jamespucyk Wrote: ------------------------------------------------------- > Time Value return calculate the monthly returns, > geometrically connect them to form quarterly, do > the same with the quarterly returns to form the > annual ones… DONE… oh and don’t annualize > partial returns. It seems you are assuming monthly returns. What is your idea on (Q1)? I need rule reference. > Internal Dispersion, can be high/low, STDev, > etc… Oh, I know this but I’d like to know what are the raw data you use to get the internal dispersion? Again, I need rule reference. - sticky

I forgot to mention that you can annualize monthly returns by geometrically linking them. It’s all in the Gips stuff man, you should read that. Your input data is your return calculations.

sticky Wrote: ------------------------------------------------------- > jamespucyk Wrote: > -------------------------------------------------- > ----- > > Time Value return calculate the monthly > returns, > > geometrically connect them to form quarterly, > do > > the same with the quarterly returns to form the > > annual ones… DONE… oh and don’t annualize > > partial returns. > > It seems you are assuming monthly returns. What > is your idea on (Q1)? I need rule reference. > > > Internal Dispersion, can be high/low, STDev, > > etc… > > Oh, I know this but I’d like to know what are the > raw data you use to get the internal dispersion? > Again, I need rule reference. > > - sticky you just do cross-sectional as oppose to time series of your composite - your data points are returns of individual portfolios

Q1, Q4, Q5 — any one? - sticky

Q1 - I don’t believe there is any specific req for p/f return calcs. The final results are published as composites so what really matters is that composite returns are calculated quarterly (till Jan 2010, then monthly). To facilitate that, P/F valuation is needed and this P/F valn is required monthly or when there is a sign CF (aft 2010). As long as the p/f valn is done, P/f return & thereafter composite can be calculated from this. - BN

BN Wrote: ------------------------------------------------------- > Q1 - I don’t believe there is any specific req for > p/f return calcs. The final results are published > as composites so what really matters is that > composite returns are calculated quarterly (till > Jan 2010, then monthly). To facilitate that, P/F > valuation is needed and this P/F valn is required > monthly or when there is a sign CF (aft 2010). As > long as the p/f valn is done, P/f return & > thereafter composite can be calculated from this. > > - BN Thanks for the insight, BN. It does clarify my confusion here. Does it mean that I even in 2010, I can simply keep monthly portfolio valuations plus valuations at large CF, and calculate NO portfolio returns during the whole year ----- and just calculate the single, annual return for each portfolio before coming up to the composite return for the year? If that’s the case, then (Q4) will be simple: you simply get each portfolio’s annual return first, based on each portfolio’s valuations during the year. Internal dispersion can be obtained from the 4 annuals returns from the 4 portfolios. Correct? Any for (Q5), any idea? - sticky