Performance Update for Colm Fagan's ARF

[JimmyB99]

Thanks @ClubMan,

The year by year returns would be fine. Ideally, someone could also calculate the cumulative returns over 1, 3, 5 and 10 years as well. This can then be the annual benchmark test.

 

[Colm Fagan]

As requested!
Attached is a spreadsheet showing returns for individual years and cumulative return over the last 15 years for my ARF and AMRF combined.
The key 15-year figures from the spreadsheet are an average time-weighted return of 10.77% pa and an average money-weighted return of 10.90% per annum.
The figures shown are actual amounts credited to my ARF/AMRF , net of all charges, the most important being the platform provider's charge, which was (I think) 0.62%pa until 2023, falling to 0.4%pa for 2024 and 2025. Charges also includes commission on share purchases and sales, stamp duty, etc. and a once-off charge from a pension consultant for advice on changing providers (not significant in the overall scheme of things).
The cash flows for 2016 and 2017 allow for a transfer in of two small insurance company ARF's.
All cash flows are assumed to have occurred mid-year. That assumption is appropriate for more recent years, but in the early years, cash flows were mainly at year end. I don't think it makes much difference.
I have some preliminary figures, which indicate that the returns compare well with those from a particular insurance company's international equity fund, and beat those from its mixed funds, invested in equities, bonds, etc.
One could argue that the international equity fund is the correct benchmark, since that's where my fund is invested. I think that a mixed fund may be a better benchmark because I have always striven for a mixture of "safe" and "risky" assets, but my definition of "safe" assets is equities paying what I think are safe high dividends. I view those as quasi-bonds.

 

[ronaldo]

That's a great reduction in charges. A lot of people wouldn't think of it this way, but a 0.22% reduction in annual charges for someone drawing down 4% of their fund annually would allow them to draw down 4.22% instead; increasing their gross income by 5.5% whilst having no impact on the fund itself.

 

[Colm Fagan]

That's a great reduction in charges.

I agree, and many thanks to @Steven Barrett for his advice.
I have an excellent relationship with the new provider and they do a great job, but I still think that the charges are too high (at least for a larger ARF) given that - in my opinion, of course - there is very little work involved.
PS: I also like how you've expressed it.

 

[Cameo]

Interesting reading as always.

I thnk a better approach to help benchmark performance would be to simulate all the returns of similarly concentrated portfolios (using the the msci world developed index as the universe).

Apparently it's easy enough to do using r or the like, for those who can

I agree 100% that charges are just too high, irish pension products looks very expensive compared to what's available in the uk

 

[Duke of Marmalade]

I thnk a better approach to help benchmark performance would be to simulate all the returns of similarly concentrated portfolios

This would show up the variability of returns but on average would reproduce the average return of the index. What would that tell us?

 

[Cameo]

This would show up the variability of returns but on average would reproduce the average return of the index. What would that tell us?

It would show how Colm's portfolio compared against all the possible options; i don't think their average returns would necessarily be the same as the index at all.

 

[JimmyB99]

Thanks everyone,

Does anyone know what global index tracker fund @GSheehy used last year so that we can prepare the comparative figures?

 

[Duke of Marmalade]

i don't think their average returns would necessarily be the same as the index at all.

The average return on a portfolio randomly selected from a universe would be the same as the average return on the universe, provided the rule for selection of the portfolio used the same weights that are used in calculating the return on the universe.
Now I doubt that Colm's selection criterion was proportional to market cap - probably more equal. In any event simulations are not needed to calculate averages.
I must be misinterpreting what you are suggesting; can you give a bit more explanation e.g. how would you propose the simulation would make its selection?

 

[Cameo]

I must be misinterpreting what you are suggesting; can you give a bit more explanation e.g. how would you propose the simulation would make its selection?

Maybe I am talking garbage but my thinking is as follows:

Colm’s approach consists of broadly of picking 10 to 15 individual stocks.

If you simulated the returns of all the possible such portfolios you would be able to get a distribution including the average and median portfolio returns.

I think such an approach is a better way to assess how the specific portfolio chosen has done

Some other constraints might be necessary E.g. setting a maximum % accounted for by an individual share

I suggested using the msci global index as a universe which I believe consists of around 2300 stocks

I don’t think the average return of such portfolios will be the sam as the index as they are much more concentrated, it would be an interesting exercise to see how close it was to the index return all the same.

I think Colm’s portfolio is very different to an msci benchmark type portfolio as its sector, country and currency splits will all be very different, unless I missed it somewhere the approach is largely unconstrained.

 

[Duke of Marmalade]

get a distribution including the average and median portfolio returns.

Well the average in the sense of the mean will be the average of the portfolio. See my post #204. I modelled a very skew universe of 10 stocks 9 of which return 2% certain and one of which (not known in advance) returns 62%. The universe of the 10 stocks will return a certain 8%. It is a mathematical truism that selecting portfolios from this universe will have the same expected mean return.
But you are right that the median does not follow this rule. For example let's say we look at portfolios of 1 of these stocks. 9 times out of 10 you will get 2% return and once you will get 62%. The median return is 2%. The median return for portfolios of 6 or more would be 62% as there is a better than 50/50 chance of picking the good guy.
I think you are also right that calculating the median in a real world scenario would be challenging analytically and would best be done using stochastic methods.
You have stimulated my interest. I am going to do some doodles on the FTSE 100 2025 performances.

 

[Duke of Marmalade]

This is the result of my first doodles. The median of a portfolio of 10 stocks is a bit smaller than the mean but not much.

I attach for information the FTSE 100 ordered by performance in 2025.

 

[Corola]

Is that 1,000 different combinations of 10 stocks randomly chosen from the FTSE 100?

I imagine the same exercise on the S&P 500 would give a wider difference between median and mean.

 

[Duke of Marmalade]

Is that 1,000 different combinations of 10 stocks randomly chosen from the FTSE 100?

Yep.

I imagine the same exercise on the S&P 500 would give a wider difference between median and mean.

I might try that.
Just to clarify if I had a portfolio of all 100 (actually I only have 99) stocks I would find no variability at all as I am doing a hindsight choice of portfolios which is not the same thing as a stochastic model of the future.
Meanwhile here are more complete doodles.

 

[Duke of Marmalade]

I managed to download the S&P 500 performance for H1 2015 (attached).
I am not very happy with the preliminary results as they suggest a quite tight distribution of portfolio returns. I will try and get full year figures. Gemini tells me that with the ticker Excel will be able to give me what I want. Anyway here is the current work in progress.

 

[Colm Fagan]

Conventional measures of riskiness are completely inappropriate for some investors. Here's why.

As readers of my investment diary know, Phoenix Group is the largest holding by far in my ARF (drawdown pension) and has been for years. The graph on the left below shows why.

Dividends have been on a steady upward trajectory for donkeys’ years, neatly matching my need for a regular income in retirement.

I have reason to believe that dividends are safe for many years to come, apart from management’s assurances to that effect.

Historically, the company’s core business has been the management of closed “with-profits” portfolios. I know – or used to know - that business.

Over two decades ago, I was With-Profits Actuary for a UK life insurer. Despite the company being closed to new business, my calculations indicated that it would continue to deliver positive cash flows for decades to come. I understand that cash flows are still positive. It is now part of Phoenix Group.

Even if there’s famine and pestilence, the business should continue to spew out cash. Actually, famine and pestilence may mean bonus dividends because of annuity holders dying prematurely!

The graph on the right below charts movements in Phoenix Group’s share price from 2019. At its lowest point, the price was down more than 35% and the dividend yield was over 10%. Experts in financial economics, who know all about Itô's lemma, Brownian motion, kurtosis, lognormal distributions, the Bessembinder effect, tell me that a share whose price fluctuates so wildly has no place in the portfolio of an old man like me: it’s far too risky.

The question of the riskiness or otherwise of shares like Phoenix Group is important in assessing the relative performance of my pension portfolio over the last 15 years, all of which were spent in drawdown.

That will be the subject of my next post.

 

[Cameo]

the Bessembinder effect,

Never heard of that one

 

[Cameo]

The question of the riskiness or otherwise of shares like Phoenix Group is important in assessing the relative performance of my pension portfolio over the last 15 years, all of which were spent in drawdown.

Your post got me thinking about trying to use a casfliow matching or liability driven approach when in drawdown phase.

You could include expected dividends from equities on the asset side.

Might be another less traditional way of setting an appropriate investment strategy when in drawdown.

 

[Fortune]

Brownian motion

Isn't that what moves the tea leaves in my teapot as it brews?

 

[Fortune]

You could include expected dividends from equities on the asset side.

I've long suspected the outcomes would be quite different between direct equity investment within the ARF and buying the usual fund in the ARF, even if it held the same equities, in the same proportions, as the direct share investment. The availability of cash dividends seems advantageous in drawdown.