measurement of risk

[Itsallaguess]

But HYP1 does not principally concern itself with capital returns.

It's the output income that matters and is a reflection of its success/failure/risk, I would suggest?

The maximum output income that you can sustain is determined by the total return and volatility.

Which I think gets to the heart of the issue as to why there's often disagreements around the word 'risk' when it comes to different investment strategies, and especially those related to the hopeful regular-delivery of dividend-based income...

You seem to be suggesting that if HYP1 had now evolved into a single ultra-large holding, then so long as any total-return and volatility metrics showed that single ultra-large HYP1 holding to be 'low risk' on your specific terms, then it would be right for it to be considered as 'low risk' by any given income-investor...

Does anyone honestly believe that an income-investor would consider that to be the case, where he might be relying on the income from that single, ultra-large HYP1 holding to continue to provide for his long-term retirement?

It's clear from the recent development of this thread that people are choosing to concentrate on risk-based metrics that fail to consider or take account of any strategic-intent of a given investment-approach, and I think that gets to the heart of the dilemma here, where the OP is attempting to square a risk-based circle only on his terms...

Cheers,

Itsallaguess

[GeoffF100]

You seem to be suggesting that if HYP1 had now evolved into a single ultra-large holding, then so long as any total-return and volatility metrics showed that single ultra-large HYP1 holding to be 'low risk' on your specific terms, then it would be right for it to be considered as 'low risk' by any given income-investor...

A quick Google search shows that even Apple has a 1 year volatility of nearly 30%. The 1 year volatility of the FTSE All-World index is about 11%. So no that will not happen. If you have a single large holding the volatility will be horrifying. Even the FTSE All-World index is too risky for most people, so they usually add bonds to dampen the risk down.

[Itsallaguess]

You seem to be suggesting that if HYP1 had now evolved into a single ultra-large holding, then so long as any total-return and volatility metrics showed that single ultra-large HYP1 holding to be 'low risk' on your specific terms, then it would be right for it to be considered as 'low risk' by any given income-investor...

A quick Google search shows that even Apple has a 1 year volatility of nearly 30%. The 1 year volatility of the FTSE All-World index is about 11%. So no that will not happen. If you have a single large holding the volatility will be horrifying. Even the FTSE All-World index is too risky for most people, so they usually add bonds to dampen the risk down.

Which, again, is completely ignoring the strategic-intent of the HYP1 strategy, or indeed, any income-oriented investment approach...

Any discussion trying to nail down 'risk' that concentrates on such metrics and fails to even consider or appreciate the importance of such strategic intent is bound to either fail, or simply go around in circles...

Cheers,

Itsallaguess

[Bubblesofearth]

You can back test. Calculate the daily total returns for a rebalanced HYP and proceed as before. Nonetheless, any result for one portfolio over one time period has its limitations.

Does it need to be daily? I appreciate that this is the approach for option pricing but for something like a HYP would longer intervals approximate the answer OK? Just thinking about feasibility here. If I did try to do a back-test then realistically yearly is about the shortest interval I'd be able to look at, i.e. where data (HYP1 for example) is easily available.

Trying to calculate for a rebalancing strategy would be very dependent on which shares were rebalanced into. This could be shares that already comprise part of the portfolio or completely new ones. The number of possibilities seems overwhelming.Or is there a theoretical short-cut?

The exercise of calculating Sharpe ratios for the different approaches is something I'd like to have a go at but only if there is a way of doing it that makes sense and is not going to take too much time!

BoE

[GeoffF100]

If you are doing the calculation over a long period, weekly data should give a reasonable indication. I do not think yearly data would be appropriate. It does not really matter how many data point you have if you download a CSV file:

https://help.yahoo.com/kb/SLN2311.html

I was not able to get sensible results for ETFs when I tried this some years ago. You really need NAV numbers rather than prices for ETFs. Individual stocks and OEICs gave sensible results though.

If you want to rebalance, I suggest annual rebalancing. Divide the total value by the number of stocks, and invest that amount in each stock.

[tjh290633]

But HYP1 does not principally concern itself with capital returns.

It's the output income that matters and is a reflection of its success/failure/risk, I would suggest?

The maximum output income that you can sustain is determined by the total return and volatility.

I think the reverse applies. Total Return depends on the income generated as a major factor.

TJH

[GeoffF100]

The maximum output income that you can sustain is determined by the total return and volatility.

I think the reverse applies. Total Return depends on the income generated as a major factor.

Total return is income plus any capital gain (or minus any capital loss). You can generate more income by selling shares.

It does not matter where the total return comes from, dividends or capital gains will do equally well. If you withdraw too much though, you will not be able to sustain your income in a downturn. Adding some bonds usually increases the amount of income that you can sustainably withdraw.

[tjh290633]

I think the reverse applies. Total Return depends on the income generated as a major factor.

Total return is income plus any capital gain (or minus any capital loss). You can generate more income by selling shares.

It does not matter where the total return comes from, dividends or capital gains will do equally well. If you withdraw too much though, you will not be able to sustain your income in a downturn. Adding some bonds usually increases the amount of income that you can sustainably withdraw.

Selling shares does not count in Total Return. Only income received and capital value changes count.

TJH

[GeoffF100]

Total return is income plus any capital gain (or minus any capital loss). You can generate more income by selling shares.

It does not matter where the total return comes from, dividends or capital gains will do equally well. If you withdraw too much though, you will not be able to sustain your income in a downturn. Adding some bonds usually increases the amount of income that you can sustainably withdraw.

Selling shares does not count in Total Return. Only income received and capital value changes count.

That is right, but you can withdraw both income received and capital gains, provided that you are not too greedy.

[Bubblesofearth]

If you are doing the calculation over a long period, weekly data should give a reasonable indication. I do not think yearly data would be appropriate. It does not really matter how many data point you have if you download a CSV file:

https://help.yahoo.com/kb/SLN2311.html

I was not able to get sensible results for ETFs when I tried this some years ago. You really need NAV numbers rather than prices for ETFs. Individual stocks and OEICs gave sensible results though.

If you want to rebalance, I suggest annual rebalancing. Divide the total value by the number of stocks, and invest that amount in each stock.

Thanks, makes sense. However I can't see myself doing this as rebalancing in the manner you describe is not something I would contemplate. The only way that I imagine it would work well is if shares mean reverted and the evidence I've seen and read about suggests that's not the case for many. My fear is that I would end up not only taking money out of star performers but would be channeling it into losers.

BoE

[GeoffF100]

If you are doing the calculation over a long period, weekly data should give a reasonable indication. I do not think yearly data would be appropriate. It does not really matter how many data point you have if you download a CSV file:

https://help.yahoo.com/kb/SLN2311.html

I was not able to get sensible results for ETFs when I tried this some years ago. You really need NAV numbers rather than prices for ETFs. Individual stocks and OEICs gave sensible results though.

If you want to rebalance, I suggest annual rebalancing. Divide the total value by the number of stocks, and invest that amount in each stock.

Thanks, makes sense. However I can't see myself doing this as rebalancing in the manner you describe is not something I would contemplate. The only way that I imagine it would work well is if shares mean reverted and the evidence I've seen and read about suggests that's not the case for many. My fear is that I would end up not only taking money out of star performers but would be channeling it into losers.

I am not suggesting that it is a good idea to do it in practice. The suggested calculation should give you some idea of the effect of rebalancing on volatility, and perhaps performance. There are lots of people here who seem to believe in equal weighting (AKA cut your winners and add to your losers). Run your winners and cut your losers? Lots of people believe in that. Evidence? I could not find anything convincing, but there are certainly momentum effects in the market. Do momentum funds always win? No. Equal weighting?

https://seekingalpha.com/article/460794 ... 00-indexes

[GoSeigen]

I think the reverse applies. Total Return depends on the income generated as a major factor.

Total return is income plus any capital gain (or minus any capital loss). You can generate more income by selling shares.

It does not matter where the total return comes from, dividends or capital gains will do equally well. If you withdraw too much though, you will not be able to sustain your income in a downturn. Adding some bonds usually increases the amount of income that you can sustainably withdraw.

Funny conversation, I expect TJH and GeoffF100 are at cross purposes due to the former's unconventional definition of "capital" and "income".


Total return is a figure derived from ALL the cash flows from an investment (as opposed to a subset of them -- only dividends to date and purchase price for example).

So a total return calculation includes purchase price, all income flows (and I would include expense flows, trading costs, tax etc.) and disposal price.

The disposal price is the tricky one. If the portfolio has been divested then it is obvious. But if the portfolio is still invested then the typical value used for disposal price is the market's valuation of all the cash flows that have not yet been received, i.e. the current market price.


No-one has ever adequately explained to me why the value of future cash flows should reasonably be excluded when evaluating the performance of an investment.'


Total return doesn't "come from" either one or the other of "income" and "capital", it derives from all the cashflows. Ultimately income and capital are an artificial distinction (the chief practical difference being the effect they have on taxation). It's easy to see this when you consider the range of government bonds in issue.


IME thinking in in more general terms is easier than creating artificial distinctions (capital/income, white/black, human/animal etc) and then trying to make sense of them.

GS

[Bubblesofearth]

I am not suggesting that it is a good idea to do it in practice. The suggested calculation should give you some idea of the effect of rebalancing on volatility, and perhaps performance. There are lots of people here who seem to believe in equal weighting (AKA cut your winners and add to your losers). Run your winners and cut your losers? Lots of people believe in that. Evidence? I could not find anything convincing, but there are certainly momentum effects in the market. Do momentum funds always win? No. Equal weighting?

https://seekingalpha.com/article/460794 ... 00-indexes

Thanks to help from itsallaguess I've managed to calculate the standard deviation for both HYP1 and the FTSE100 over the 22 year lifetime of HYP1. I've only used annual data as that's all that is readily available so I accept the inherent inaccuracy this will entail. The result is that HYP1 has a SD some 2.68x that shown by the index.

I've not attempted any SD calculations for rebalanced portfolios as I'm not comfortable doing this whilst retaining all the original shares. It's not a method by which I would invest.

One thought that occurs to me is that SD calculations would, by their nature, give a volatility for cash on deposit that is higher than cash under the mattress, i.e. it would be deemed higher risk. The higher the interest rate the higher this SD, and therefore volatility risk, would be. Indeed, if the interest rate were high enough it could give an SD higher than the stock market. This is clearly nonsense as cash on deposit is generally seen as zero risk.

Does this mean that the SD for a more strongly growing portfolio will tend to be higher than for a stagnant one? Is this a reason HYP1 shows a higher SD than FTSE100? If so then the attribution of higher risk for HYP1 could be wrong in any sense that matters to the investor.

Far me to argue with various Nobel laureates but I do wonder if there is a better measure of risk than SD?

On the subject of investment strategy my preference is for equal weight on purchase followed by LTBH. Outside of HYP there is very little in the literature about this approach, probably because it's not one that easily lends itself to fund management.

BoE

[torata]

I am not suggesting that it is a good idea to do it in practice. The suggested calculation should give you some idea of the effect of rebalancing on volatility, and perhaps performance. There are lots of people here who seem to believe in equal weighting (AKA cut your winners and add to your losers). Run your winners and cut your losers? Lots of people believe in that. Evidence? I could not find anything convincing, but there are certainly momentum effects in the market. Do momentum funds always win? No. Equal weighting?

https://seekingalpha.com/article/460794 ... 00-indexes

Thanks to help from itsallaguess I've managed to calculate the standard deviation for both HYP1 and the FTSE100 over the 22 year lifetime of HYP1. I've only used annual data as that's all that is readily available so I accept the inherent inaccuracy this will entail. The result is that HYP1 has a SD some 2.68x that shown by the index.

I've not attempted any SD calculations for rebalanced portfolios as I'm not comfortable doing this whilst retaining all the original shares. It's not a method by which I would invest.

One thought that occurs to me is that SD calculations would, by their nature, give a volatility for cash on deposit that is higher than cash under the mattress, i.e. it would be deemed higher risk. The higher the interest rate the higher this SD, and therefore volatility risk, would be. Indeed, if the interest rate were high enough it could give an SD higher than the stock market. This is clearly nonsense as cash on deposit is generally seen as zero risk.

Does this mean that the SD for a more strongly growing portfolio will tend to be higher than for a stagnant one? Is this a reason HYP1 shows a higher SD than FTSE100? If so then the attribution of higher risk for HYP1 could be wrong in any sense that matters to the investor.

Far me to argue with various Nobel laureates but I do wonder if there is a better measure of risk than SD?

On the subject of investment strategy my preference is for equal weight on purchase followed by LTBH. Outside of HYP there is very little in the literature about this approach, probably because it's not one that easily lends itself to fund management.

BoE

Hi BoE
Possibly I've completely misunderstood what you've done, but surely the SD is (in my layman's language) the width of a bell curve spread of the individual components so that 1SD covers roughly 68% of the components. I can understand why it might be interesting to see how HYP SD changes over time from being initially equally weighted, but I don't see how you can draw any conclusions from comparing it to the FTSE 100, nor why you think that a bigger SD could mean stronger growth.
This is not any kind of criticism, I simply don't understand the reasoning.
torata

Edit: I think you can ignore my post. Thinking about it, I guess you have done the calculation on the returns for each of 21 years? I assumed it was the SD of the portfolio of individual stocks over time.

[GeoffF100]

One thought that occurs to me is that SD calculations would, by their nature, give a volatility for cash on deposit that is higher than cash under the mattress, i.e. it would be deemed higher risk.

No. The volatility of cash earning a fixed rate of interest is zero. Here is the correct calculation for volatility:

https://www.macroption.com/historical-volatility-excel/

Suppose cash earns 5% p.a. First we need to calculate Ri = ln (Ci / Ci-1). (The i and i-1 are subscripts.) This will be the same every day.

Ci / Ci-1 = 1.05^(1/364)
Ri = ln(1.05^(1/364))

The standard deviation of Ri over any time period is zero. As I have said, yearly prices are not good enough. You can get daily prices from Yahoo and other sources. The free sources tend to contain "noise" though, i.e. daily movements that did not occur in the real market.

[Bubblesofearth]

No. The volatility of cash earning a fixed rate of interest is zero. Here is the correct calculation for volatility:

https://www.macroption.com/historical-volatility-excel/

Suppose cash earns 5% p.a. First we need to calculate Ri = ln (Ci / Ci-1). (The i and i-1 are subscripts.) This will be the same every day.

Ci / Ci-1 = 1.05^(1/364)
Ri = ln(1.05^(1/364))

The standard deviation of Ri over any time period is zero. As I have said, yearly prices are not good enough. You can get daily prices from Yahoo and other sources. The free sources tend to contain "noise" though, i.e. daily movements that did not occur in the real market.

This is testing the limits of my maths understanding but I think I see my mistake. I calculated SD based on differences from the mean of annual closing prices rather than differences in ln annual returns from the mean annual return.

Let me know if that's correct and I'll recalculate for my own curiosity. I appreciate annual data is not really good enough.

BoE

[Urbandreamer]

One thought that occurs to me is that SD calculations would, by their nature, give a volatility for cash on deposit that is higher than cash under the mattress, i.e. it would be deemed higher risk.

No. The volatility of cash earning a fixed rate of interest is zero. Here is the correct calculation for volatility:

https://www.macroption.com/historical-volatility-excel/

I just need a quick clarification of that statement.
Obviously that formula will provide the volatility of the price of gold, a given share, some random crypto or even the price of bread.
What do you price cash relative to in order to calculate it's volatility?

Were I to price £'s in bitcoin, I bet that we would clearly see that the £ is very volatile, though I would hazard a guess that everyone on TLF would argue that the volatility is not that of the £ but that of bitcoin.

So, is it actually true, in any meaningful sense, to claim that cash in savings has zero volatility, if it is measured against itself?

[GeoffF100]

No. The volatility of cash earning a fixed rate of interest is zero. Here is the correct calculation for volatility:

https://www.macroption.com/historical-volatility-excel/

Suppose cash earns 5% p.a. First we need to calculate Ri = ln (Ci / Ci-1). (The i and i-1 are subscripts.) This will be the same every day.

Ci / Ci-1 = 1.05^(1/364)
Ri = ln(1.05^(1/364))

The standard deviation of Ri over any time period is zero. As I have said, yearly prices are not good enough. You can get daily prices from Yahoo and other sources. The free sources tend to contain "noise" though, i.e. daily movements that did not occur in the real market.

This is testing the limits of my maths understanding but I think I see my mistake. I calculated SD based on differences from the mean of annual closing prices rather than differences in ln annual returns from the mean annual return.

No, they are both wrong. Follow the example in the link. For each year, divide the price at the end of the year by the price at the start of the year. Take the natural logarithm or the result for each year. Calculate the standard deviation of these results. Divide by the square root of 1, which is 1.

[GeoffF100]

No. The volatility of cash earning a fixed rate of interest is zero. Here is the correct calculation for volatility:

https://www.macroption.com/historical-volatility-excel/

I just need a quick clarification of that statement.
Obviously that formula will provide the volatility of the price of gold, a given share, some random crypto or even the price of bread.
What do you price cash relative to in order to calculate it's volatility?

Were I to price £'s in bitcoin, I bet that we would clearly see that the £ is very volatile, though I would hazard a guess that everyone on TLF would argue that the volatility is not that of the £ but that of bitcoin.

So, is it actually true, in any meaningful sense, to claim that cash in savings has zero volatility, if it is measured against itself?

Suppose that you are interested in the return after the passage of yen years. A gilt maturing in ten years will give a known return in GBP. The volatility of that return (a measure of its variability) will be zero (assuming that the UK government does not default on its debt). If you by a US Treasury Bond maturing in ten years, you will receive a known return in USD at the end of that ten years. The volatility of that return in USD will be zero. The volatility of that return in GBP will be the volatility of the GBP/USD exchange rate. UK investors usually hedge foreign bonds into sterling, which is usually the currency of their liabilities.

If you invest your money in a series of over-night sterling bank deposits, your return over ten years will not be certain. You can calculate the volatility over the previous ten years to give an indication of the likely variability of the return (the annual volatility multiplied by the square root of 10).

[vand]

Risk is not just about volatility, but anyone who says that volatility does not mean risk is stupid and misinformed. Volatility is certainly a form of risk - otherwise we would all be investing in black swan funds that happily losing 80%/year for 19 years in the assuredness that the one time it does pay off will net you enough to pay it all back.

Risk, whatever your preferred metric, and return are ying and yang. You can't have excess return without taking on investment risk, and its a incomplete to consider one of them without also considering the other. "Risk adjusted return" is a concept everyone here should be informed on, imo.

We take on uncertainty in equities for the possibility of higher returns than are available in cash and fixed income - the common forms of passive investment. You hope that your stocks will beat bonds over the next 5, 10, 20 years, but you know what? that is never guaranteed.