|
Posted by David on April 9, 2008, 2:31 am
. =A0Would you like to take a stab at it?
>
> A prediction for the Dow at the =A0end of 2008? Of course. No problem.
> But bear in mind that a proper prediction has an expected value and a
> standard deviation (SD)!
>
> Checking with Riegels (Third Edition) for the historical statistics
> over 1946-2001, he gives the average annual Stock Market capital
> appreciation, without dividends, as 7.5% with an SD of 16.2%. Scaling
> this for the remaining nine months or three quarters of a year,
> linearly for return and as the square root for the SD, gives an
> expected gain of 5.6% with an SD of 14.0%.
>
> This means there is a 66% probability that the change from the present
> value to the end of 2008 will be in the range +19.6% to -8.4% (1 SD)
> and a 95% probability that it will be in the range +33.6% to -22.4% (2
> SDs).
>
> The present value of the Dow is 12,576.4 so the predicted ranges are
>
> Expected value =3D 13,281
>
> 66% probability it will be between 11,520 and 15,041
>
> 95% probability it will be between 9,759 and 16,802.
>
> So, a value of 15,000 is quite possible as it is only about 1 SD up
> from the expected value, but then 11,520 is just as likely and 13,281
> is the most likely.
>
> This assumes that stock market returns follow a random walk with trend
> and that the logs of changes follow a normal distribution. This is not
> quite true of course as the distribution of returns is not really
> Gaussian but tends to follow Hirst statistics which have fatter tails
> than the normal distrivbution so that extreme changes are more
> likely.
>
> But, sticking with a normal distribution for simplicity, a Dow of
> 20,000 would be 3.8 SDs above the expected value, which is possible
> but unlikely, with a probability of less than 0.1%
>
> A quick check on the change in the Dow over the last year shows a
> +0.1% gain, which is just 0.46 SDs down on the expected value, so it
> is all consistent and as expected.
>
> This is not as exciting as your article's claims of "Dow 15,000 by end
> of year" but rather more scientific and likely to be true.
>
>
>
>
>
> > >http://time-blog.com/real_clear_politics/2008/04/dow_15000.html
>
> > Hmmm... The article is going along quite well until he says that his
> > forecasting method uses fundamentals including the interest rate and
> > that the interest rate is now so low that the method predicts a Dow of
> > 20,000! So, as he thinks this is the wrong answer, he uses a 10-year
> > treasury yield of 6% instead, to get a prediction of 15,000.
>
> > This means that instead of a predicted rise of 7,500 from the present
> > 12,500 to 20,000 he arbitrarily changes the parameter used in his
> > formula to get a rise of 2,500 or 1/3rd as much! Is this how you do
> > economics in the USA now :-)
>
> > Are we supposed to believe this stuff? It tends to make the
> > traditional examination of chicken's entrails or the casting of
> > horoscopes look quite repectable and scientific.- Hide quoted text -
>
> - Show quoted text -
Actually I should have said "Siegel" instead of "Riegels". I am
confusing the eminent professor with a long-dead German
aerodynamicist. Tsk, tsk!
| 
Yahoo! 
Google 
Windows Live 
del.icio.us 
digg 
Netscape 
XML Sitemap