Re: stat measures

On 9 Aug 2005 16:19:03 -0700, "kaferro@xxxxxxxxxxx"
<kaferro@xxxxxxxxxxx> wrote:

>
>I am designing a system to test some basic investing ideas, such as
>what happens after a 10% market drop or four down days in a row. A
>famous author suggested that one should test a trading idea with a
>study that:
>1. enumerates all observations of what happened after a defined market
>event over a specific period of time;
>2. weighs whether the results indicate a random phenomenon or a
>tradable anomaly;
>3. measures the uncertainty associated with the later conclusion; and
>4. predicts the probability that an x-percent move will follow the
>event being studied.
>
>Can you suggest the statistical measures for #2,#3, and #4. I have
>begun to study statistics. In the meantime, though, since I am
>currently writing the program, your suggestions will allow me to code
>the stats now as I continue the process of learning about statistics in
>general.
>
>Thank you

You might want to take a look at Burton Malkiel's A Random Walk Down
Wall Street. For a more technical look, try Lo and MacKinlay, A
Non-Random Walk Down Wall Street.

As others have pointed out, to a first approximation stock prices in
widely traded companies follow a random walk with drift. This is also
true to a second approximation.

You aren't going to get a simple answer to the questions you posed
because there isn't a single, simple answer. The behavior of
statistics based on nonstationary series, such as stock prices. can be
"nonstandard."

If you like to program, here's something to mess around with. First,
program up the ability to simulate random series. Then generate a
bunch of such series and count up how often you observe a move of
x-percent (or whatever). Plot a histogram. Then take the real data and
see where it's characteristics fall in histogram of data that you know
is unpredictable.

By the way, you can make a good living doing financial econometrics,
but it generally takes five or six years of graduate study.
-Dick Startz
----------------------
Richard Startz RichardStartz@xxxxxxxxxxx
Lundberg Startz Associates
.

Relevant Pages

  • Re: Minimum Structural Complexity and Gap Distance - a clear relationship
    ... Neutral changes cannot be guided by natural selection Howard. ... Random walk does not have any statistical ... involve long random walks in sequence space. ... You need to learn a bit more about the statistics of blind searches. ...
    (talk.origins)
  • Re: stat measures
    ... > what happens after a 10% market drop or four down days in a row. ... Use you money to buy a monkey, ... throw darts at the stocks page to pick the stocks for you. ... > begun to study statistics. ...
    (sci.stat.math)
  • Re: Delphi in the right direction
    ... I have serious doubts about the validity of any statistics developed based on popular search engines, at least as long as we don't know how they structured their queries. ... Since China and India have over 3 times the population of the US, you would think that logic would dictate that these are the markets you would want target: ... China: 1.3 Billion potential customers/consumers ... While raw numbers of market potential are important, the ability of the customer/consumer to purchase goods and service is even more important. ...
    (borland.public.delphi.non-technical)
  • Re: Microsoft jscript / mshtml coding standards
    ... now as low as 50% to 60%, depending on whose statistics you believe. ... the credibility of a numerical analysis such as statistics should be ... which says the *total* market share of Internet Explorer ... As IE runs only in Windows, that would mean that of every 9 PC's ...
    (comp.lang.javascript)
  • Re: Market Share vs Installed Base
    ... Analysts and reporters like to talk about market share statistics, ... market share of the Zune among iPods, ... Lies, Damn Lies, and then there's Statistics! ... being late on any replacement cycle. ...
    (comp.sys.mac.advocacy)