The Lemon Fool

I've been tracking my portfolio returns using both XIRR & unitisation (the accumulation model). Would someone mind explaining (in layman's terms ) what the differences between these 2 numbers tells me please?

As my understanding goes:

• XIRR gives a money-weighted return: so time periods in which money is invested have more of an impact on the overall return than equivalent time periods when less money is invested.
• Unitisation gives you a time-weighted return: it tells you the underlying investment performance and strips out the impact of money flowing in & out.

Currently, my unitisation number is at about 14%, my XIRR is closer to 8%. The benchmark I'm using is at around 10%.

I'm trying to understand how to interpret the numbers: does the higher unitisation % mean that the timing of my investments has had an overall negative impact? I'm also not sure which number to believe - as in, have I achieved a 14% return, or is it actually 8%?

Also, the benchmark I'm using is VWRL (I couldn't find a suitable index, so figured I should at least make sure I was beating an all-world ETF as if not, I should probably just put my money into that instead...) Should I be comparing my XIRR or unitisation number against the benchmark though?

Many thanks,
iambic.

iambic wrote:As my understanding goes:

• XIRR gives a money-weighted return: so time periods in which money is invested have more of an impact on the overall return than equivalent time periods when less money is invested.
• Unitisation gives you a time-weighted return: it tells you the underlying investment performance and strips out the impact of money flowing in & out.

I'm not sure I like the "money-weighted" and "time-weighted" labels, but putting that aside, yes, you've got it. XIRR includes the effects of cash flows and unitisation strips it out.

does the higher unitisation % mean that the timing of my investments has had an overall negative impact? I'm also not sure which number to believe - as in, have I achieved a 14% return, or is it actually 8%?

Yes. Both; they measure different things.

Also, the benchmark I'm using is VWRL (I couldn't find a suitable index, so figured I should at least make sure I was beating an all-world ETF as if not, I should probably just put my money into that instead...) Should I be comparing my XIRR or unitisation number against the benchmark though?

VWRL is a distributing ETF and so if you're going to use that you'll need to compare against your porfolio's income units. VWRP is the accumulation version.

Comparison can be done directly against your unitisation numbers. You can also compare against your XIRR figure if you calculate a parallel XIRR figure for the benchmark; i.e. simulate having invested the same £ amounts on the same dates into the benchmark as into your portfolio.

iambic wrote:Currently, my unitisation number is at about 14%, my XIRR is closer to 8%. The benchmark I'm using is at around 10%.

I assume you've annualised your Acc price performance? If acc price has gone from 1.00 to 1.14 in 2 years and your IRR shows as 8% then you would be comparing 2 years performance with 1 for IRR.

I am puzzled. I unitise on both models and get exactly the same result from XIRR from the cash flow on both. The OP must be doing something wrong in his calculations. If the dividends are reinvested in his portfolio, then they do not feature in his cash flow.

TJH

tjh290633 wrote:I am puzzled. I unitise on both models and get exactly the same result from XIRR from the cash flow on both.

The OP is comparing return as measured by accumulation units vs return as measured by XIRR, not the XIRR on acc vs inc units.

mc2fool wrote:
VWRL is a distributing ETF and so if you're going to use that you'll need to compare against your porfolio's income units. VWRP is the accumulation version.

Comparison can be done directly against your unitisation numbers. You can also compare against your XIRR figure if you calculate a parallel XIRR figure for the benchmark; i.e. simulate having invested the same £ amounts on the same dates into the benchmark as into your portfolio.

Thanks very much for clarifying & also for pointing me in the direction of VWRP. I've updated the benchamrk numbers now (it looks like VWRP only started back in June 2019) & it is currently sitting at 3.12% (as opposed to VWRL at around 8%), so it's nice to see my return look a little more positive.

nmdhqbc wrote:I assume you've annualised your Acc price performance? If acc price has gone from 1.00 to 1.14 in 2 years and your IRR shows as 8% then you would be comparing 2 years performance with 1 for IRR.

No I haven't annualised my unitised performance. Sorry for the newbie question, but should I have?

I've unitised right from the start of my investing & have done the same with the XIRR (so all inflows & outflows from the start date are used in the calculations), so I thought they would be calculating over the same period of time?

iambic wrote:I've unitised right from the start of my investing & have done the same with the XIRR (so all inflows & outflows from the start date are used in the calculations), so I thought they would be calculating over the same period of time?

XIRR is quoted as an annualised rate,so anything with which you compare it should be annualised as well.

iambic wrote:No I haven't annualised my unitised performance. Sorry for the newbie question, but should I have?

Ah, well that makes a difference. Assuming you are using Excel, and assuming you have your 14% unitised gain in cell A1 as 0.14, then use:

=POWER(1+A1,1/years)-1

For years put in a reference to a cell that contains: =(lastdatecell-firstdatecell)/365.25

Alaric wrote:XIRR is quoted as an annualised rate,so anything with which you compare it should be annualised as well.

Ah I see, thanks for clarifying.

mc2fool wrote:Ah, well that makes a difference. Assuming you are using Excel, and assuming you have your 14% unitised gain in cell A1 as 0.14, then use:

=POWER(1+A1,1/years)-1

For years put in a reference to a cell that contains: =(lastdatecell-firstdatecell)/365.25

Thanks for the clear explanation; I've added this to my tracking spreadsheet & it turns out my annualised return is 5.24%. So that's a lot closer to the 8% XIRR & I think also means my timing has actually had an overall positive impact? (More down to luck than judgment but still interesting to see & good to know my timing wasn't as disastrous as the previous 8% vs 14% suggested :p )

In general, since both acc units and portfolio XIRR are attempting to get you a view of your overall Total Return you would expect the numbers to converge as portfolios mature.

As I have posted elsewhere we run a portfolio system on the cloud(originally developed by my late father in law( OZYU), I did make a small contribution and run the thing, and train users, these days), and currently I have read access 35 portfolios belonging to family and close friends, unitised identically for acc and inc units, plus portfolio XIRR. The oldest is from 1982(my mother in law's taxed portfolio, although she was an investor way before that). For every portfolio older than 7 years old, the convergence is very good indeed. But please see the caveats below.

We always start acc units at £1, because it is then easy to plot acc unit progress on the same graph as (1+XIRR)^years. (or as said above you could plot the acc unit compound return directly vs XIRR over the same time period).

Caveat, Don't expect convergence in the following cases.

Portfolios less than one year old comparison meaningless unless you 'correct' XIRR for less than full year timespan, even then poor convergence usually.

Young portfolios just a few years old, mostly poorish convergence due to the fact that cash movements can be a large relative proportion of overall portfolio value in these early years.

Any portfolio, of any age, with very large (violent) cash movements in relation to portfolio value. ie Ernie smiled on you and you fed £1m into a £250k portfolio. This would send XIRR in a spin for a while, but in a year or so convergence would resume.


Example: Our own three portfolios, two PEP/ISAs since beginning of 1987 and our taxed portfolio since April 1993 all show very close convergence, such that the data plots of Acc units and (1+XIRR)^duration are basically undistinguishable. For example, at last night's close, in our taxed portfolio Acc units have compounded at 12.47% since inception, portfolio XIRR is 12.48%. If you don't get this convergence with no violent cash flows, simple, your unitisation is faulty.

Bagger

mc2fool wrote:

tjh290633 wrote:I am puzzled. I unitise on both models and get exactly the same result from XIRR from the cash flow on both.

The OP is comparing return as measured by accumulation units vs return as measured by XIRR, not the XIRR on acc vs inc units.

I see what you mean. To get an IRR for the unit value you need the first and last value and dates, which takes no account of the creation and elimination of units, and doing that I get an XIRR of 10.05%, as opposed to 8.98% for the cash flow. The Income units give 5.01%, but that takes no account of units created, from dividends and cash inputs, or destroyed when cash is withrawn.

In my case both units began at unity on 21 April 1987 and are respectively £24.01 and £5.06 tonight.

TJH