measurement of risk

[Alaric]

That is based on the assumption that it is equally desirable to hold the stock or to hold a risk free bond.

In practice investing institutions are constrained by mandates which require them to hold one of equities or bonds, but not both. Didn't the formulae for pricing options break during financial crises? If the derivation of the formulae contains assumptions which aren't always valid, that' quite possible.

[GeoffF100]

In fact, while the causes of the volatility can't be predicted, they are NOT random. There ARE causes.

Volatility is assumed to be random in the models. Of course there are causes for what actually happens, but they cannot be predicted in advance. The options market is perhaps the best evidence that the models are statistically sound. Option traders have some refinements to make the Black-Scholes formula more accurate. The "volatility smile" is one.

If you thought that a full recovery was inevitable after the Covid crash, you could have bet on it with the futures market directly or via a spread bet, but you would have lost your shirt if you were wrong. The implied volatility would have soured after the Covid crash, so call options would have been expensive, particularly if you only had a vague idea of when the recovery would happen.

There was only about a 15% drop in the all world index in GBP terms. I made good use of it nonetheless.

[GeoffF100]

That is based on the assumption that it is equally desirable to hold the stock or to hold a risk free bond.

In practice investing institutions are constrained by mandates which require them to hold one of equities or bonds, but not both. Didn't the formulae for pricing options break during financial crises? If the derivation of the formulae contains assumptions which aren't always valid, that' quite possible.

Market makers in the option market need to hold both, so that they can delta hedge. My text book is rather old now. The options market sailed through nicely in the 1987 flash crash, which was more precipitous than anything that is allowed nowadays. Portfolio insurance was gutted.

[Steveam]

The reason this is important is to minimise misunderstandings. If a definition is too vague, or not quantified, then it opens itself to disagreement without any yardstick to say who is right or wrong. For example, during one discussion around HYP1, some people regard the imbalance as too risky whilst others accept it. Without a clear definition of exactly what the risk is and how to quantify it you can end up in a situation where both parties feel they are right even if diametrically opposed. You can also end up going round in endless circles!

I think you have to have a common agreement on what the yardstick is first, before you get into those questions.

I don't think many people would disagree that the minimum acceptable performance of HYP1 would be to preserve the spending power of that initial £3,451. 22 years later that would be £6,776 (I am using RPI/CHAW here https://www.ons.gov.uk/economy/inflatio ... /chaw/mm23). HYP1 produced £11,123 that year, RIO contributing ~27%. Does that make it "risky" ? The yardstick would suggest not at all, as indeed this year's income collapse to £7,729 has shown - it is still ahead.
And yet in year 3, with a more equally balanced portfolio which certainly didn't look as "risky" as it might look now, its income fell 8%, well short of yardstick.

All risks then are relative. But relative to what?

Marks suggests (and I think he’s correct) that you can’t assess risk in retrospect. Your outcome is just one of the many possible outcomes (some having higher probability than others). I refer to the paper linked by simoan above.

Best wishes,

Steve

[Bubblesofearth]

Marks suggests (and I think he’s correct) that you can’t assess risk in retrospect. Your outcome is just one of the many possible outcomes (some having higher probability than others). I refer to the paper linked by simoan above.

Best wishes,

Steve

Which kind of takes us to the question as to whether volatility is a good measure of risk. If risk cannot be assessed by the past then volatility becomes a poor measure. But at the same time future outcomes are unknowable and therefore risk becomes unquantifiable. So where does that leave us in trying to compare the risk of different investment strategies?

If we go back to HYP1 as an example we can ask whether it is riskier to leave alone or rebalance periodically. If you leave alone you run the risk of the portfolio becoming unbalanced. If you rebalance you run the risk of missing out on the full effect of any big winners that might emerge. Whether deemed negligible or not, rebalancing also incurs transaction costs.

If we cannot quantify the above risks then how do we decide which is the greater?

BoE

[Steveam]

I’m going to withdraw from this thread. Marks addresses many of the points being raised but people are either not reading what he has written or are being wilful in misunderstanding. I’m not going to try to paraphrase or summarise his memo.

BoE: you’ve just said “risk cannot be assessed by the past” but this isn’t what I said, nor what I meant. I think you are reading “you can’t assess risk in retrospect” and summarising it as “risk cannot be assessed by the past” - I don’t think they are the same but my wording was not that of an expert or deep thinker on this subject. Go back to the original memo.

Geoff100: you castigate Marks for his understanding of volatility but he carefully discusses what he means by volatility. You may be disagreeing with his definition and usage but he’s quite clear and on his definition and usage (which he argues is that of the ordinary investor) he is right. You’re moving the goalposts and then criticising.

Best wishes,

Steve

[GeoffF100]

Marks suggests (and I think he’s correct) that you can’t assess risk in retrospect. Your outcome is just one of the many possible outcomes (some having higher probability than others). I refer to the paper linked by simoan above.

Which kind of takes us to the question as to whether volatility is a good measure of risk. If risk cannot be assessed by the past then volatility becomes a poor measure.

Actually, risk can be assessed by the past. A simple rule is that if you are pricing an option that expires in n months, you calculate the volatility over the previous n months. (There are more complicated methods that are a little more accurate statistically.) We do have a clue what will actually happen (active investors will think otherwise, of course). We cannot assess the outcome in advance, but we can assess the risk.

Let us revisit the situation at what we now know was the bottom of the Covid crash. Because of a mathematical result called put call parity, the cost of an at the money put option would have been the same as the price of an at the money call option with the same expiration date. The cost of betting on a fall in the market would have been the same as the cost of betting on a rise in the market.

[Urbandreamer]

Which kind of takes us to the question as to whether volatility is a good measure of risk. If risk cannot be assessed by the past then volatility becomes a poor measure.

Actually, risk can be assessed by the past. A simple rule is that if you are pricing an option that expires in n months, you calculate the volatility over the previous n months. (There are more complicated methods that are a little more accurate statistically.) We do have a clue what will actually happen (active investors will think otherwise, of course). We cannot assess the outcome in advance, but we can assess the risk.

I could be totally off base, but are you talking about what Long Term Capital Management did?
https://www.investopedia.com/terms/l/lo ... apital.asp

The model that you keep referring to is still one of the best mathematical models we have. But it's a model and limited to what can be expressed easily using mathematics. It tends to ignore real world events and risks associated with them, assuming that they will be reflected in volatility or price. That assumption is not always valid.

I'd also disagree that it's always impossible to assess outcomes in advance. I made money spread betting upon HMV, because the future was obvious, though unpopular.
Mr Soros, claimed no difficulty in assessing the outcome of the BoE's support of the pound during the ERM crisis. There are times that the outcome is fairly obvious.

[Bubblesofearth]

I’m going to withdraw from this thread. Marks addresses many of the points being raised but people are either not reading what he has written or are being wilful in misunderstanding. I’m not going to try to paraphrase or summarise his memo.

BoE: you’ve just said “risk cannot be assessed by the past” but this isn’t what I said, nor what I meant. I think you are reading “you can’t assess risk in retrospect” and summarising it as “risk cannot be assessed by the past” - I don’t think they are the same but my wording was not that of an expert or deep thinker on this subject. Go back to the original memo.

Geoff100: you castigate Marks for his understanding of volatility but he carefully discusses what he means by volatility. You may be disagreeing with his definition and usage but he’s quite clear and on his definition and usage (which he argues is that of the ordinary investor) he is right. You’re moving the goalposts and then criticising.

Best wishes,

Steve

The definition of 'in retrospect' is 'with hindsight' or 'looking back on the past'

https://www.google.com/search?client=sa ... 8&oe=UTF-8

Marks opening lines state that volatility falls far short as the definition of investment risk. He goes on to say that what investors fear is permanent loss of capital which, unlike volatility, (he states) is not quantifiable. Although I don't entirely agree with this I do have some sympathy with it. The GFC threw up what at the time were described as 6 sigma events, i.e events that were 6 standard deviations away from the norm and therefore so extremely unlikely that you would not have predicted them from looking at variance as occurring more than once every 1.38 million years. This is of course where Taleb comes in.

It's a shame if you withdraw from the debate as IMO this is an important topic both from the pov of investment decisions and for clarity of discussions around risk. The fact that it's a complex subject means that there will inevitably be disagreements and misunderstandings along the way. The ultimate objective is to further our understanding. That at least was the intention of my original question.

BoE

[GeoffF100]

I could be totally off base, but are you talking about what Long Term Capital Management did?
https://www.investopedia.com/terms/l/lo ... apital.asp

No.

[GeoffF100]

The GFC threw up what at the time were described as 6 sigma events, i.e events that were 6 standard deviations away from the norm and therefore so extremely unlikely that you would not have predicted them from looking at variance as occurring more than once every 1.38 million years.

Everyone knows that the log normal distribution of stock prices only an approximation, so they correct for that. In practice the distribution has "fat tails". I have already referred to the volatility smile:

https://en.wikipedia.org/wiki/Volatility_smile

As you will see, the statistical modelling is becoming ever more sophisticated.

[GeoffF100]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

[Bubblesofearth]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

OK but how does that help me decide which strategy (rebalancing vs LTBH) to adopt?

BoE

[moorfield]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

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?

[Arborbridge]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

This seems irrelevant to me. I'm not concerned about volatility since this is of virtually no importance. My risk is a long term one: will my income cease or suffer existentially? will my capital tend to zero between now and when I stop depending on it?
Nothin else matters, so I'm not faffing around worrying about volatility.

Arb.

[GeoffF100]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

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.

[Arborbridge]

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.

Maybe so, but short term volatility is not relevant here. No one gives a damn about the wobbles: only about the general direction of travel.

[GeoffF100]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

This seems irrelevant to me. I'm not concerned about volatility since this is of virtually no importance. My risk is a long term one: will my income cease or suffer existentially? will my capital tend to zero between now and when I stop depending on it?
Nothin else matters, so I'm not faffing around worrying about volatility.

You may not care, but others will care about their capital and the maximum income that they can sustain. If we want to make comparisons with other approaches, we need to use a standard yardstick.

[GeoffF100]

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

Maybe so, but short term volatility is not relevant here. No one gives a damn about the wobbles: only about the general direction of travel.

The volatility that is most relevant here is the long term volatility over the life of the HYP so far. The higher the volatility the more likely it is that you will either go bust or not be able to withdraw much income.

[GeoffF100]

As far as the risk of HYP1 is concerned, a standard procedure is:

(1). Get noise free daily closing prices for the stocks concerned.
(2). Work out the daily total returns.
(3). Calculate the volatility of the total return over the period concerned.
(4). Compare that with the volatility of the all-world total return index over the same period.

If you want to know how well HYP1 has done, calculate the Sharpe ratio for HYP1 and compare it with the Sharpe ratio for the all-world total return index.

OK but how does that help me decide which strategy (rebalancing vs LTBH) to adopt?

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.