Crawling Road blog post - Permanent Portfolio Performance and Historical Returns

[hedgehog]

I guess I asked the same question before but it remains unanswered so here I go again:
https://web.archive.org/web/20160324133 ... l-returns/

The question: if I recall correctly, some risk measures were featured in this post, like
- Standard deviation
- Sharpe(?)

but strangely, they where removed.

Why? Are risk measures not important? Or is it just my memory that plays games with me?

Anyways, I am still interested in the average Standard Deviation and Sharpe for the Permanent Portfolio on a historical period, thanks.

[nwagers]

I don't remember the Sharpe ratio being on there, but this might help:

http://mebfaber.com/2013/07/31/asset-al ... ategies-2/

[gizmo_rat]

Simba's backtesting excel spreadsheet has more portfolio (inc PP) ratio's, charting and comparisons than a sane person could need.

If you can think of it it'll do it, I seem to remember it was pretty straightforward too.

http://www.bogleheads.org/forum/viewtop ... 4#p1912664

[Gosso]

The CAGR in the following table is inflation-adjusted and the standard deviation is based on monthly returns (and then adjusted to reflect annual stdev).  The portfolio is rebalanced monthly, transactions are frictionless, and dividends/interest are reinvested.  I assumed a T-Bill rate of zero for the sharpe ratio since I'm working with inflation-adjusted numbers.

[dragoncar]

I don't think it was ever there:

https://web.archive.org/web/20090302210 ... l-returns/

[hedgehog]

I don't remember the Sharpe ratio being on there, but this might help:

http://mebfaber.com/2013/07/31/asset-al ... ategies-2/

I very much like this graph. Thanks for bringing it to my attention. What can be concluded of it?

Comparing CARG, Sharpe and Standard Deviation, is my assumption correct that the best of the pack is the "Risk Parity Portfolio"?

[craigr]

I guess I asked the same question before but it remains unanswered so here I go again:
https://web.archive.org/web/20160324133 ... l-returns/

The question: if I recall correctly, some risk measures were featured in this post, like
- Standard deviation
- Sharpe(?)

but strangely, they where removed.

Why? Are risk measures not important? Or is it just my memory that plays games with me?

Anyways, I am still interested in the average Standard Deviation and Sharpe for the Permanent Portfolio on a historical period, thanks.

I don't think I put Sharpe/SD in that post. I'm not a fan of Sharpe/SD as they are misleading in terms of actual risk. And mostly I don't talk much about these risk measures because they are rearward looking and sometimes project a false sense of security.

Over time the common measures tend to miss really dangerous fat-tail risks that can cause a lot of portfolio damage. I think the common risk measures can provide some guidance to how a portfolio has done under some extreme scenarios in the past, but can't really do much in terms of telling you what will happen in the future. In other words, I think they are useful for at least eliminating bad ideas that have previously failed even if they can't really be used to show what will work going forward.

[hedgehog]

I don't think I put Sharpe/SD in that post. I'm not a fan of Sharpe/SD as they are misleading in terms of actual risk. And mostly I don't talk much about these risk measures because they are rearward looking and sometimes project a false sense of security.

Over time the common measures tend to miss really dangerous fat-tail risks that can cause a lot of portfolio damage. I think the common risk measures can provide some guidance to how a portfolio has done under some extreme scenarios in the past, but can't really do much in terms of telling you what will happen in the future. In other words, I think they are useful for at least eliminating bad ideas that have previously failed even if they can't really be used to show what will work going forward.

Now comes the obvious question: what are good ways to measure risk then, if not Standard Deviation/Sharpe?

Vanguard has a nice series (which I haven't read yet):
https://personal.vanguard.com/us/insigh ... principles

[dragoncar]

I don't think I put Sharpe/SD in that post. I'm not a fan of Sharpe/SD as they are misleading in terms of actual risk. And mostly I don't talk much about these risk measures because they are rearward looking and sometimes project a false sense of security.

Over time the common measures tend to miss really dangerous fat-tail risks that can cause a lot of portfolio damage. I think the common risk measures can provide some guidance to how a portfolio has done under some extreme scenarios in the past, but can't really do much in terms of telling you what will happen in the future. In other words, I think they are useful for at least eliminating bad ideas that have previously failed even if they can't really be used to show what will work going forward.

Now comes the obvious question: what are good ways to measure risk then, if not Standard Deviation/Sharpe?

Vanguard has a nice series (which I haven't read yet):
https://personal.vanguard.com/us/insigh ... principles

I think the problem is measuring past volatility is not the same as predicting future risk

[craigr]

I don't think I put Sharpe/SD in that post. I'm not a fan of Sharpe/SD as they are misleading in terms of actual risk. And mostly I don't talk much about these risk measures because they are rearward looking and sometimes project a false sense of security.

Over time the common measures tend to miss really dangerous fat-tail risks that can cause a lot of portfolio damage. I think the common risk measures can provide some guidance to how a portfolio has done under some extreme scenarios in the past, but can't really do much in terms of telling you what will happen in the future. In other words, I think they are useful for at least eliminating bad ideas that have previously failed even if they can't really be used to show what will work going forward.

Now comes the obvious question: what are good ways to measure risk then, if not Standard Deviation/Sharpe?

Vanguard has a nice series (which I haven't read yet):
https://personal.vanguard.com/us/insigh ... principles

I think this starts getting pretty theoretical and how you view risk and uncertainty.

From my view again things like SD/Sharpe can only show you what worked in the past over that particular history in that particular country. It really tells you nothing about what happens when things go off the rails. And unfortunately in human history things are always going off the rails far more than they are in a state of stable bliss.

So in those terms, I think it's better to have options to deal with risks than various scenarios and measures to try to predict anything. For instance, I hold gold because history has shown currency debasement and collapse to be a very common risk in paper money. I don't really care what jingoism and national flag is wrapped around it. It's best to diversify. Same thing for owning stocks. Traditionally stocks have weathered some pretty bad economic events even if the ride has been bumpy. Then there is risk of war, government collapse, civil strife, natural disasters, etc.

None of these things can be quantified with std. deviation and really it's a fool's errand trying to do so. So there needs to be a mix of looking at the past for sure and throwing out stuff that obviously failed. Then there is the other side which is protecting even against very extreme situations because they do happen, and likely will happen at least once over an investor's lifetime and possibly more than once.

For me, one of the biggest mistakes an investor can make is assuming the past will repeat into the future. Unfortunately this is the kind of thinking std. deviation and Sharpe ratio promote (same with CAGR numbers). That's why I get tired of stock bug arguments that stocks always win over time so you should own a lot of them. The data used to arrive to this conclusion is very short, and very focused (mainly the U.S.).  When you break these two parameters stocks are actually quite risky. You have to diversify well beyond stocks to be really sure you are protected against the future. 

[hedgehog]

Thank you button.

For me, one of the biggest mistakes an investor can make is assuming the past will repeat into the future.

Ain't it the case that history repeats itself (think wars) if we don't learn from it?

[craigr]

Thank you button.

For me, one of the biggest mistakes an investor can make is assuming the past will repeat into the future.

Ain't it the case that history repeats itself (think wars) if we don't learn from it?

When has humanity ever learned from history enough to stop the next war?

[gap]

I agree it is a mistake for investors to assume history will repeat. However, I think it is another mistake to ignore history -there aren't too many "laws of investing" we can use or trust.  Diversification may be the only one but that, too, is difficult to implement perfectly because strictly speaking  it involves predicting the future based on past behavior of assets.

Personally, when I drive, I do look at the rear view mirror and the side mirror and anything else that might give me a clue