Rich get richer as economy gets better; everyone else is worse off

[Malthus]

We are talking about why there exists a gap or disperity between how well the upper-whatever percent have done since the recession, and the rest.

Not sure where you are getting 'Diversification doesn't really "ensure" high return' from when I never said it did.

You said that the rich had done better (i.e. got higher return) because they were diversified, and that this result was predictable (i.e. ensured).  You are right, you didn't say that diversification ensures higher returns, you just said something that is entirely equivalent to it.

Well, except for the fact that I said nothing like either part of what you have for some reason attributed to me, you are totally correct. 

Look, if you are of the opinion that "predictable" is the equivalent of "ensured", there is really no help for you.

Do you think that the statement "if you drink and drive, getting into an accident is an entirely predictable result" really means "if you drink and drive, getting into an accident is ensured"?

Isn't your profession supposed to be about risks and stuff? 

[DGuller]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

[crazy canuck]

Isn't your profession supposed to be about risks and stuff? 

[DGuller]

Isn't your profession supposed to be about risks and stuff? 

I don't see what so in it.  This discussion seems to be about semantic bullshit, not risk, which is what Malthus's profession is supposed to be about.

[fhdz]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

[Malthus]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

What you are missing is the context. In this analogy, the "accident" has already happened, the car is in the ditch with wheels spinning - that is, we already know that the non-rich have fared more poorly.

While the chances of getting into any accident may be only 1% for any particular trip, some rather higher figure of accidents are "caused or contributed" to by drinking and driving. So if you drink and drive and get into an accident, yes, it's "predictable" that the two would be related, right? While by no means being "ensured"?   

I'd love to see you arguing otherwise in court: "yes, your honour, my client was drunk as a freaking skunk when he plowed into that minivan full of nuns, but there was no way he could reasonably have predicted his drinking would cause or contribute to that - why look at the figures, only 1% of drunk trips result in accidents!" 

[Malthus]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

[derspiess]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

[fhdz]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

The small comfort I feel in the fact that I would likely not pass the actuarial exams is the reassurance that I can at least parse everyday language reasonably well.

[DGuller]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

I didn't take it to be guaranteed either, hence the use of quotes around "ensured".  For brevity, I did not go into all the exceptions that are necessarily part of the financial theory.  However, it's even not statistically likely, in any meaningful interpretation of the term, that a diversified portfolio would outperform an undiversified portfolio.  The advantages of diversification do not directly lead to superior returns, they're more subtle than that.

[DGuller]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

What you are missing is the context. In this analogy, the "accident" has already happened, the car is in the ditch with wheels spinning - that is, we already know that the non-rich have fared more poorly.

While the chances of getting into any accident may be only 1% for any particular trip, some rather higher figure of accidents are "caused or contributed" to by drinking and driving. So if you drink and drive and get into an accident, yes, it's "predictable" that the two would be related, right? While by no means being "ensured"?   

I'd love to see you arguing otherwise in court: "yes, your honour, my client was drunk as a freaking skunk when he plowed into that minivan full of nuns, but there was no way he could reasonably have predicted his drinking would cause or contribute to that - why look at the figures, only 1% of drunk trips result in accidents!" 

Well, everything is predictable after it already happened.

[fhdz]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

I didn't take it to be guaranteed either, hence the use of quotes around "ensured".  For brevity, I did not go into all the exceptions that are necessarily part of the financial theory.  However, it's even not statistically likely, in any meaningful interpretation of the term, that a diversified portfolio would outperform an undiversified portfolio.  The advantages of diversification do not directly lead to superior returns, they're more subtle than that.

It seems to me that the main advantage of diversification is that it minimizes the risk of loss rather than that it maximizes the chance of return.

[DGuller]

It seems to me that the main advantage of diversification is that it minimizes the risk of loss rather than that it maximizes the chance of return.

It minimizes both risk of abnormal loss and risk of abnormal gains, among other things.  If you want to shoot for a highest return, like for example in a stock picking context, you want to be as un-diversified as possible.  Your best chance of winning is to put all your money into a company that's about bankrupt, hoping that it would rebound from near-zero stock price.  That's why such contests are so stupid.

[fhdz]

It seems to me that the main advantage of diversification is that it minimizes the risk of loss rather than that it maximizes the chance of return.

It minimizes both risk of loss and risk of gains, among other things.

Yes, I think that's what I was trying to say.

Having said that, if your assets are large to begin with, smaller gains over a longer period of time might be preferable to the chance that a huge investment will ruin you.

Guller, what's your opinion of an investment strategy which would look something like 90% of assets in very low-risk investment and 10% in ridiculously risky ventures? (Or 80/20, or whatever split might give one the best spread between lowest potential downside and large potential upside)

[Malthus]

Well, the chance of getting into an accident if you drive drunk is below 1%, but let's assume it's exactly 1%.  So, if at least 1% qualifies as "predictable" in your world, then of course your statement was correct.  Pretty much any result is predictable.

I think what he meant by predictable was "statistically likely", not "guaranteed".

I didn't take it to be guaranteed either, hence the use of quotes around "ensured".  For brevity, I did not go into all the exceptions that are necessarily part of the financial theory.  However, it's even not statistically likely, in any meaningful interpretation of the term, that a diversified portfolio would outperform an undiversified portfolio.  The advantages of diversification do not directly lead to superior returns, they're more subtle than that.

There is nothing subtle about this.

The "accident has already happened" - one asset class has, to put it mildly, had "issues".

Diversification is all about preventing shit like that from happening.

http://www.investopedia.com/articles/02/111502.asp

Thus, if you get all screwed up because you invested in one asset class alone, this is an "entirely predictable" result - much like a drunk driver plowing into a vanful of nuns ought, in hindsight, to have known that drinking and driving was a bad, bad idea. Drinking and driving is a bad idea because it increases your risks. Investing in one asset is a bad idea for much the same reason.

Now, no doubt you will go on about how, actually, plowing into a vanful of nuns is massively unlikely (I mean, how many nuns are there around anyway?) but this is to miss the larger point.