bostonbubble.com

[admin]

Thanks, you too.

[Renting in Mass]

[GenXer]

Multiplicative cascades in action. Multi-sigma moves in the market are nothing new. Trying to figure out why or how is useless. When a bunch of players all jump in at the same time, that's what you get. Does it mean much? Not really. Just like in the Stock Market, you get behavior like that a lot more often than people realize. Just because something was 'biggest on record' means nothing - we can have moves in the future that are even bigger - for example, the move downward in the next couple of years could be biggest ever on record. Volatility clusters, and this is what we are going to see sooner or later (sooner, more likely).

As far as investment, the more volatility you got, the less appealing the investment from a long-term point of view. Stocks actually returned anything from a full loss to multi-100% gains. S&P500 index (if we are talking about 'the market') returned roughly 8% over the past 20 years, but the benefit investors saw is minimal from good stock performance unless you actually held exactly the S&P index in your portfolio. Same way, Case-Schiller may have returned 7.75% over 20 years, but the benefit, again, is quite minimal, unless you happened to hold a big house portfolio. Individual houses did better or worse over that time, and I actually would say that individual houses probably did worse because the longer you 'hold' - the more likely you are to spend more on repairs taxes,etc, expenses which grew much faster than inflation.

A side note: median is a useless measure when the mean is so far away from median. This is a sign that the statistics of your distribution is fat-tailed. For such distribution, mean and median (as well as standard deviation) are relatively poor descriptions of the data, because mean is different from median (sometimes significantly), and because standard deviation can vary so much that it becomes just another moving part, therefore containing no useful information. You may want to look at skew (3rd moment) and kurtosis (4th moment), but even these higher moments may be moving especially during times of big changes and big price moves.

[Renting in Mass]

I looked at a Pearson Type VII distribution and found that the leptokurtic densities have a higher peak than the mesokurtic normal density. I think we all know what that means...

[GenXer]

It means that the real tails are fatter, and nothing more

It also means that the probability of small moves is a bit larger (<1 sigma), probability of medium moves is a bit smaller and probability of large moves is much larger. But this is true only if you compare two distributions side by side, and given that you know what they are. There is no one distribution that describes the markets. So this information is unfortunately academic. What is not academic is the 'fat tail' behavior -the multi sigma jumps that basically tell us we're dealing with a leptokurtic distribution. And the fact ath small moves happen a bit more often is another reason why most people are simply fooled by randomness into believing that everything is Normal.