How Unbalanced is your HYP?

[funduffer]

In revamping my HYP as I described in the 10 year review I posted recently (viewtopic.php?f=15&t=42330), I got to thinking how much effort should I put into balancing up the capital and income from the portfolio. Now, some HYPers here aim for quite a strict balance using top slicing, topping up or adding capital to keep capital and income from each portfolio constituent within strict limits. Others, including pyad with the famous HYP1, allow the portfolio to go wherever it takes itself, leading to quite unbalanced contributions from each constituent over time.

So, is there a measure of how unbalanced a portfolio is? I decided to create one myself! I introduce to you two ideas of how to characterise your portfolio – the ‘fundiagram’, and the ‘funbalance coefficient’ (patents pending, copyright applied for!).

Let us use HYP1, as described in the Nov 2023 review to illustrate these. (viewtopic.php?f=15&t=41459).

HYP1 is often criticised for being very unbalanced and I expect that all portfolios left to their own devices over a 20+ year period will become like HYP1. So, let us look at the fundiagram and funbalance coefficient for HYP1. These can be calculated for both capital and income, but let us start with income. (Exactly the same process can be used for capital).

We start by writing down the portfolio in descending order of annual income, then calculate the % of income. Then we accumulated these percentages. We then calculate the cumulative percentages from a perfectly balanced portfolio with the same number of constituents. Finally we subtract the balanced cumulative percentages from the HYP1 cumulative percentages, sum them up and divide by the number of constituents and multiply by 2. This gives us the funbalance coefficient! This will vary between 0% for a perfectly balanced portfolio, to 100% for one where just one constituent provides all the income.

Here are the details for HYP1:

HYP1 Income



This produces an income funbalance coefficient of 54%. You can interpret this to mean it is roughly halfway between a perfectly balanced portfolio and one where just 1 share provides all the income, and the others provide none!

The position can be shown visually on a fundiagram:



I have added my own portfolio (FD HYP) to this chart, which lies between HYP1 and the perfectly balanced portfolio. The income funbalance coefficient for FD HYP is 30%.

The fundiagram for capital looks like this:




I have added a line for tjh’s portfolio as reported recently. (viewtopic.php?f=15&t=25337)

The capital funbalance coefficient for HYP1 is 47%, for FD HYP it is 30% and for tjh’s HYP it is 10%, reflecting the top slice / top up system he operates.

You may ask, how is this useful?

I would argue that it is useful for comparing portfolios of different sizes and numbers of constituents. It could also be used to monitor how imbalanced a portfolio becomes over time, and perhaps being useful to consider rebalancing when a chosen threshold is reached.

For me, if the funbalance coefficient exceeds 25%, I start to get uncomfortable about concentration risk, so I might do something, as I am about to with my HYP. Others, like pyad, probably see this all as a waste of time, and would accept any value!

Anyway, I hope I haven’t bored you too much.

Fun question for the mathematicians amongst you – how many shares would you need to own to get a funbalance coefficient of exactly 100%?


FD

[daveh]

That's going to take me ages with 45 constituents plus I'll do it for the 2023 calendar year as that is how my data is arranged, but I'll give it a go and see what it comes up with, but no graphs for me as I'm useless (I was going to use a word beginning with c but assume the nanny filter will stop me) at posting images.

[tjh290633]

Thank you for including my data on your chart. However I should point out that embedded charts are supposed to be accompanied by the URL where they are hosted, plus an assurance that they are your own intellectual policy. This is not in doubt in this case, but it is important that the site is protected as indicated in the site rules.

Obviously the coefficient is dependent on the extent to which imbalance is tolerated. I allow 50% above the median holding value, but with a lower value portfolio 100% could be justified. I do not think that tighter limits could be used practically

See above

TJH

[moorfield]

There's no analysis of capital weights by your coefficient here, is that intentional?

An exaggerated example, if you have instead one share contributing zero income but holding 85% of overall value, with the other 15 holding the remaining 15% and each contributing 6.67%, your coefficient tells us that is a very nearly perfectly balanced portfolio, which is nonsense clearly. So its a flawed method (and dangerous one - if you are going to use it to start making buying/selling decisions you will sooner or later make a mistake you didn't need to). Absolute and arbitrary limits also - why 25%? Why not 26%, 27%, or 29%? You can see the rabbit holes that opens... and ultimately its your decision making that suffers. Such measurements ideally should be relative ones. (Relative to what though, for another thread...)

It's an interesting exercise certainly (and recced), but I wouldn't use it I'm afraid. Reminds me of something I wrote here once: The most dangerous thing a HYPster can do to their hard earned and saved is be bored with it. Perhaps that's what's really going on!

[funduffer]

There's no analysis of capital weights by your coefficient here, is that intentional?

An exaggerated example, if you have instead one share contributing zero income but holding 85% of overall value, with the other 15 holding the remaining 15% and each contributing 6.67%, your coefficient tells us that is a very nearly perfectly balanced portfolio, which is nonsense clearly. So its a flawed method (and dangerous one - if you are going to use it to start making buying/selling decisions you will sooner or later make a mistake you didn't need to). Absolute and arbitrary limits also - why 25%? Why not 26%, 27%, or 29%? You can see the rabbit holes that opens... and ultimately its your decision making that suffers. Such measurements ideally should be relative ones. (Relative to what though, for another thread...)

It's an interesting exercise certainly (and recced), but I wouldn't use it I'm afraid. Reminds me of something I wrote here once: The most dangerous thing a HYPster can do to their hard earned and saved is be bored with it. Perhaps that's what's really going on!

There are 2 unbalance coefficients that you can calculate - one for income and one for capital value.

In your example, the one for capital would tell you that your portfolio was unbalanced.

As regards a threshold - that is a personal matter related to one’s own risk tolerance.

Hope this helps.

FD

[Arborbridge]

What fun!

But whether an investor prefers to have a balanced portfolio seems a matter of taste - no one has really shown how critical it is.

I expect if you did my HYP it would be similar to TJH but slightly less balanced as you could predict from my looser adherence to trimming.


Arb.

[idpickering]

What fun!

But whether an investor prefers to have a balanced portfolio seems a matter of taste - no one has really shown how critical it is.

I expect if you did my HYP it would be similar to TJH but slightly less balanced as you could predict from my looser adherence to trimming.


Arb.

Fair comments Arb. On your first one I believe PYAD used to advice topping a share up to average capital value weighting. That was mentioned in his HYP Savers article, which I can't seem to find a link to? I tend to do that, ie topping up in that manner, save for LGEN which is of double value weighting in my HYP, but let's not stray off topic.

Ian.

[Hypster]

On your first one I believe PYAD used to advice topping a share up to average capital value weighting. That was mentioned in his HYP Savers article, which I can't seem to find a link to?
Ian.

https://web.archive.org/web/20071231131 ... avers.aspx

[kempiejon]

Mine is very unbalanced.
I'm not much of one for data mining my holdings. I doubt I'll be building a new spreadsheet for my funbalance coefficient. Perhaps next winter when the long nights come.
A quick run down the HYPTUSS list and I see my largest income amount is hundreds of times larger than my smallest. That piddler is clearly an outlier, spun out of a bigger plc. Stripping that out and there's still incomes 10s of times bigger than the smallest. My shares have been accumulated rather lumpily and capital values are similarly unbalanced. My build started at low monthly investments of about £300 and some shares quickly ruled themselves out, my monthly investments quickly multiplied, better candidates evolved so top ups never came and rump holdings were never sold. I occasionally consider a sweeping tidy up but frankly don't care. Perhaps when I start taking income; if I were to sell down ISA holdings to top up my pensions they would be first to go.

[idpickering]

Mine is very unbalanced.
I'm not much of one for data mining my holdings. I doubt I'll be building a new spreadsheet for my funbalance coefficient. Perhaps next winter when the long nights come.
A quick run down the HYPTUSS list and I see my largest income amount is hundreds of times larger than my smallest. That piddler is clearly an outlier, spun out of a bigger plc. Stripping that out and there's still incomes 10s of times bigger than the smallest. My shares have been accumulated rather lumpily and capital values are similarly unbalanced. My build started at low monthly investments of about £300 and some shares quickly ruled themselves out, my monthly investments quickly multiplied, better candidates evolved so top ups never came and rump holdings were never sold. I occasionally consider a sweeping tidy up but frankly don't care. Perhaps when I start taking income; if I were to sell down ISA holdings to top up my pensions they would be first to go.

Thanks for your input. To be honest, I find your attitude to this matter very refreshing, especially hereabouts. But each to their own of course.

Ian.

[Newroad]

Hi All.

It seems to me that for this conversation to be meaningful, people need to say

What are their goals, and
What are their constraints?

No constraints and maximum income irrespective of risk will give a different "balancing" strategy than, say, mine. For me, it's pretty much

1. Equal number of shares from each of the eleven sectors
2. Highest yielding share(s) from each sector drawn from the FTSE100
3. "Optimised" weighting of each share chosen by (1) & (2) above based on risk, using volatility as a proxy

As a matter of theory, if not practise, this should keep a fairly high and stable income in context. However (3) is not "Pyadic", not KIS - so may not be for some/many. Others may have different reasons for disliking it, or finding it unsuitable - fair enough.

Regards, Newroad

[Charlottesquare]

And I thought it was to be a simple approach for Doris.

Afraid whilst I appreciate/understand unbalanced portfolio potential risk I balance mine more by seat of the pants juggling rather than science/calculation. I also do not just have single company shares in mine but a mix of these with ITs (I know, mentioned on the Practical Board) and the ITs get to be a more significant percentage of all holdings by value , the largest of these is a world trust being over 10% by value of the total.

I struggle with the point of all this analysis, it is either a simple buy and forget approach relying on income or it is not, if not, if monitoring of reliance on individual shares in needed ,then frankly allowing ITs is the really simple way to get that diversification that even Doris could manage.

[idpickering]

On your first one I believe PYAD used to advice topping a share up to average capital value weighting. That was mentioned in his HYP Savers article, which I can't seem to find a link to?
Ian.

https://web.archive.org/web/20071231131 ... avers.aspx

Thank you. I used to refer back to this item often. That’s why I’ve doubled up in my LGEN shares to be honest. The higher yielding shares in my 24 share HYP will be added to in this manner going forward too, each in turn, unless a good looking newbie appears that is, and that doesn’t happen all that often maybe?

Ian.

[daveh]

That's going to take me ages with 45 constituents plus I'll do it for the 2023 calendar year as that is how my data is arranged, but I'll give it a go and see what it comes up with, but no graphs for me as I'm useless (I was going to use a word beginning with c but assume the nanny filter will stop me) at posting images.

I get a coefficient of 35% for income for 2023 income. Maximum is 5% for anyone share. A balanced portfolio would be 2.222, but ~ half are above that and ~ half below with 5 each producing less than 10% of the balance income figure of 2.2recuring.

Not sure I'm, I mean I'm sure I'm not going to do anything with it except perhaps cull some of the low dividend producers, though some are being held until they are needed for the capital loss.