prediction methods

[Dan Goodman]

What prediction method or methods (if any) do you consider useful for
predicting the future?

Methods I've seen include:
See what newspapers give most space to. If true, "There's a new pet in
the White House" and "one sports team has beaten another" would be among
the most important signs of the future.

There are regular historical cycles.

Astrology.

See what the experts predict, and assume the opposite will happen.

--
Dan Goodman

[SolomonW]

Forgetting astrology
Essentially all methods work on two systems.
1) Taking historical examples and extrapolating
a) Historian methods an example here would be British tank theorist in the
1920s, who looked at Mongol military tactics to determine what a future
wars would look like.
b) Statistical methods
Looking at historical numbers to see if one can determine a pattern. A very
good site that does this well is
http://fivethirtyeight.blogs.nytimes.com/

What happens for example to determine the next election, you take voting
figures in the last 20 elections, add as much financial, political, etc
figures are you can get. Then put it into some data mining program and see
what happens. A very good program for this would be formulize
http://formulize.nutonian.com/

2) Listening to experts and taking a view from that.

The fundamental belief here is that the future will follow patterns from
the past. Does it?

Compare for example your views of the tourist trade in Florida and Cuba in
the next five years and you get a feel of the problem.

[David Tenner]

Dan Goodman <dsg...@iphouse.com> wrote in
news:iY-dnfB2KZHstaDM...@iphouse.net :

Allan Lichtman with his "thirteen keys to the White House" system
http://en.wikipedia.org/wiki/The_Keys_to_the_White_House has correctly
predicted the winner of the popular vote in every US presidential election
since 1984. (That might not be so impressive if he had done so, say, a week
before each presidential election, but he predicted Reagan's re-election in
1983 and Obama's in 2010, when neither was considered a sure thing.) The
keys have been attacked for being too subjective (how do we know whether a
candidate is "charismatic" or a scandal is "major", etc.) but *as interpreted
by Lichtman* they seem to work.

Of course eventually he will probably be wrong.

--
David Tenner
dte...@ameritech.net

[Dan Goodman]

Sort of; sometimes.

And, of course, sometimes the experts say things have changed forever.

> Compare for example your views of the tourist trade in Florida and Cuba
> in the next five years and you get a feel of the problem.

A while back, I read the US Forest Service's longish-range forecast of
North American forest resources. They gave three sets of predictions;
high, low, and medium.

And then cautioned that this was all assuming no major surprises in
demographics, technology, the economy, etc.

Oh -- and the forecast was for the US and Canada. Leaving out St. Pierre
and Miquelon might not make much difference; but Mexico?



--
Dan Goodman

[Dan Goodman]

Thanks. I'll have to look that up!



--
Dan Goodman

[SolomonW]

One major issue is that people learn for example the statesmen working out
a post ww2, looked at the problems at the end of ww1.

[SolomonW]

On Mon, 11 Mar 2013 10:48:55 -0500, David Tenner wrote:

He already has been. Allan Lichtman stated that Gore would be president.
Gore won the popular vote but he did not become president.

[David Tenner]

SolomonW <Solo...@citi.com> wrote in
news:vsbr2r946bs3$.8k13hbzb...@40tude.net :

I thought he explicitly said that the method only applied to the popular
vote. After all, the keys were first designed as a system not just to
predict future elections but to explain past ones (1860 through 1980), and
that included 1876 and 1888, when the winner of the popular vote did not get
win in the Electoral College.

In fact, a 1996 review makes this point: "The highest plurality of the
popular vote, not the electoral vote which actually decides the presidency,
is the criterion...[This] means that in two elections, 1876 and 1888, the
'winner' as defined in this method did not end up as president."
http://www.orms-today.org/orms-10-96/white-house.html

--
David Tenner
dte...@ameritech.net

[SolomonW]

Well he stated that Gore would be president.


http://www.questia.com/library/1G1-67412616/election-2000-the-keys-point-to-gore

"Based on the performance of the Clinton administration, Vice President Al
Gore will be the next president of the United States. There is little or
nothing that challenging-party candidate Texas Governor George W. Bush can
do to revive his presidential prospects. "

[David Tenner]

SolomonW <Solo...@citi.com> wrote in
news:10sl0aank8opc.b...@40tude.net :

OK, then, evidently he assumed that the winner of the popular vote would
also probably win the electoral vote and that 1876 and 1888 were rare
exceptions unlikely to recur. But while he was wrong about that, the fact
remains that the keys method (as he interpreted it) did get the winner of
the popular vote right, and that was my point: what I said was that using
it he "has correctly predicted the winner of the popular vote in every US
presidential election since 1984." Very likely someday in some close
election his method will be wrong about the popular vote, too, but it
hasn't happened yet.

OTOH, it is questionable whether the keys method really tells us much we
couldn't know from simpler models. This is Jonathan Bernstein's critique:

"What he has is a combination of things that are generally causes (such as
the economy) of incumbent party success, things that are effects of
incumbent party success (such as the incumbent winning renomination
uncontested and third party challenges), and things that are arbitrary and
dubious (such as whether the candidates have 'charisma').

"It's not surprising that you can 'predict' the winner with that batch of
stuff. After all, while Lichtman's system has worked since he debuted it
for the 1984 cycle, a much simpler system that predicts the incumbent party
wins barring an election-year recession also successfully calls the winners
from 1984 through 2008, at least if you count 2000 for Gore (as Lichtman
does). That doesn't make Lichtman wrong as much as it just means his system
isn't telling us much that we don't already know otherwise."

http://plainblogaboutpolitics.blogspot.com/2011/08/not-buying-keys.html

(I'm not sure whether either Lichtman's model--which makes an election-year
recession one of the keys--nor Bernstein's simpler model--which makes it
*the* determining factor--actually works for 1992, when the recession was,
at least technically, over. I suppose the answer is that people
*perceived* the economy as still being in recession in 1992. But that was
true in 2012 as well, at least as late as April. http://tinyurl.com/bl8s8fc
Anyway, Bernstein's simpler model is open to the objection that unlike
Lichtman's it is definitely wrong about some *pre-* 1984 elections, e.g.,
1952 and 1968.)

--
David Tenner
dte...@ameritech.net

[SolomonW]

His system is a propriety system that he owns, one key point of any system
is that two people working independently need to be able to come up with
the same result. In this case even the creator got it wrong.

> the fact
> remains that the keys method (as he interpreted it) did get the winner of
> the popular vote right, and that was my point: what I said was that using
> it he "has correctly predicted the winner of the popular vote in every US
> presidential election since 1984." Very likely someday in some close
> election his method will be wrong about the popular vote, too, but it
> hasn't happened yet.

It is of very limited interest who wins the popular vote, what people are
interested in is who will win the elections. Obviously there is more then
just this system required to do that.