Bespoke Portfolio vs VWRL or HMWO etc

dunstonh

research is 1 thing. but measurement is a bit spurious when it comes designing portfolios. because that is based on measuring past volatilities and co-variances of components of the portfolio, and assuming they'll be the same in the future. but volatilities and co-variances change all the time. saying "25% in X looks about right" is just as good as saying "22.78% in X because that's what the software tells me".

Variations of a few oercent here and there are not the issue and weightings change based on analysis and opinion. I was thinking more about some of the "research" that people have done which looks more like 10% into 10 funds that appeared in the Telegraph or Mail money sections. A basket full of fashion investing.

bostonerimus

ArchBair wrote: »

We are 100% equities in both of our portfolio's mainly because we prefer to have a very decent cash buffer in preference to investing in bonds/gilts etc

Cash should be included in your asset allocation. VWRL will certainly give you broad exposure to the world stock markets at minimal cost, but historically the potential extra returns of a 100% equity portfolio
over say a 60/40 portfolio are thought by many people not to justify the extra risk and volatility. Of course today and the future will probably be different from the past, but how different is the question.

Linton

grey_gym_sock wrote: »

.......
research is 1 thing. but measurement is a bit spurious when it comes designing portfolios. because that is based on measuring past volatilities and co-variances of components of the portfolio, and assuming they'll be the same in the future. but volatilities and co-variances change all the time. saying "25% in X looks about right" is just as good as saying "22.78% in X because that's what the software tells me".

Yes, worrying about a 1% or 2% difference in allocations probably isnt worth the effort, but without any measurement or some level of analysis you are just as likely to say that 45% in X feels right.

Co-variance is useful in driving an investment strategy. For example the co-variance data one can get from morningstar shows quite clearly that a set of large company funds in different geographies show a much higher correlation than a set of small company funds. One may feel this should be so but its reassuring to see actual numbers. Similarly, or perhaps as a result of the covariance numbers, the Trustnet risk score of a portfolio of small company funds can be lower than that for a global tracker. Perhaps more of a surprise.

grey_gym_sock

Linton wrote: »

Co-variance is useful in driving an investment strategy. For example the co-variance data one can get from morningstar shows quite clearly that a set of large company funds in different geographies show a much higher correlation than a set of small company funds. One may feel this should be so but its reassuring to see actual numbers.

i agree it can be nice to see the numbers. the thing is to put the numbers together with an explanation - e.g. the large companies do business more globally, so you would expect them to be more similar than smaller companies, which (on average) operate more locally. certainly, the numbers can give you more confidence in the explanation.

it's using the numbers without explanation that can be very dodgy. e.g. if you try to optimize your regional allocations for risk vs return, based on past correlations, the pure numbers often seems to tell you that, if you want to go higher risk / higher return, then you should put a large percentage in pacific ex-japan. which looks like data mining, based on a period in which that region happened to do well.

Similarly, or perhaps as a result of the covariance numbers, the Trustnet risk score of a portfolio of small company funds can be lower than that for a global tracker. Perhaps more of a surprise.

i think that one's debatable. you might come up with explanations for why that might be true. but OTOH, perhaps it's an accident of the numbers. volatility is (as well as being variable) not the same thing as risk. so i'd treat the "risk score" as food for thought.

bostonerimus

Linton wrote: »

Yes, worrying about a 1% or 2% difference in allocations probably isnt worth the effort, but without any measurement or some level of analysis you are just as likely to say that 45% in X feels right.

Co-variance is useful in driving an investment strategy. For example the co-variance data one can get from morningstar shows quite clearly that a set of large company funds in different geographies show a much higher correlation than a set of small company funds. One may feel this should be so but its reassuring to see actual numbers. Similarly, or perhaps as a result of the covariance numbers, the Trustnet risk score of a portfolio of small company funds can be lower than that for a global tracker. Perhaps more of a surprise.

Systematic differences between the past, present and future and single unknowable events are going to mess up forecasts that rely on historical data. However you can look at the statistics of historical returns and get an idea of the range of performance you might expect and the relative probability of certain outcomes if the data and your investing horizon takes in enough investment cycles and we assume that any systematic differences are not large......a communist coup in the US which nationalized all capital might be an issue for the models.

Of course in any data set it's often easy to find what you want. I've worked on data from astronomy to molecular biology and one constant is the ability of researchers to find patterns that confirm their theories or to extrapolate a result from a small statistical sample.

dunstonh

i think that one's debatable. you might come up with explanations for why that might be true. but OTOH, perhaps it's an accident of the numbers. volatility is (as well as being variable) not the same thing as risk. so i'd treat the "risk score" as food for thought.

The FE score includes a diversification measure. The paid for version has a measure and it will tell you the quality (as they see it) of the diversification. I used it today as part of a fund switch for a new client who had a bunch of multi-asset funds. The diversification measure was just 2% (i.e. barely worth the effort and possibly doing more harm than good).

So, when measuring multiple funds with little overlap, the FE risk score can lower through decent diversification.

I can see a lot of merit with the FE risk scoring method but, like most methods, it has quirks. Some funds can move around a lot within the FE score range and you need to be aware of the movement range that is likely and not just the current snapshot. Yet those that use it tend to look at the snapshot of there and then.

For example, VLS40 has moved up the FE risk scale quite significantly over the last year. Whereas L&GMI4 has remained relatively steady. A cautious investor who was going by a single snapshot on FE risk a year ago may be surprised at the increase in level of risk that the fund has had. The movement is not unexpected as VLS is return focused and not risk targeted (whereas L&GMI4 is risk targeted).

No tool or measurement system is perfect and quirks exist on all. There are over 200 data fields for underlying assets. Some risk measurement tools accept as little as 10 in their measurement. So they will map the 200 fields into 10. That can create a greater margin of error than one that has far more.

ArchBair

bostonerimus wrote: »

Cash should be included in your asset allocation. VWRL will certainly give you broad exposure to the world stock markets at minimal cost, but historically the potential extra returns of a 100% equity portfolio
over say a 60/40 portfolio are thought by many people not to justify the extra risk and volatility. Of course today and the future will probably be different from the past, but how different is the question.

So with the cash that we hold our portfolio's are 80/20. I still feel my bespoke portfolio will eventually do better than the two trackers my wife holds so it will be interesting to see what happens in a few more years time and as Bowlhead said overall we have a blend of the two.

Personally, I don't agree that historically the returns of a 100 per cent equity portfolio does not just the risk/volatility of a 60/40 portfolio.

TheTracker

ArchBair wrote: »

Personally, I don't agree that historically the returns of a 100 per cent equity portfolio does not just the risk/volatility of a 60/40 portfolio.

Yes, it’s a “personal” decision. Ultimately, at some point, *everyone* decides not to trade x% of volatility for y% of long term return.

On that basis, I’m interested that you stopped at 100%. You sound quite comfortable with it. Do you feel the returns of a 120 per cent equity portfolio do not justify the risk/volatility versus a 100% equity portfolio?