Perfect forecast of SP500 since 2 months

I searched for the most auto correlated SPX data with the last 82 days and found the period coming from august 1980

Since 2 months it has perfectly shown the way

We will test it today to see if SP500 makes a 3 - 5% move up in the next week

http://groups.yahoo.com/group/futuresystem/

I am allways amazed to find charts like that and more amazed that they produce such terrific result
 

Attachments

So you're running a time series analysis based solely on price. Mostly like differencing of degree 1, i.e. a pure momentum play?

I am a bit confused ... you're extrapolating solely on ONE period? I.E. you're using 1980 data to predict 2003?? Data mining? Over-optimization?


Quote from fortuna:

I searched for the most auto correlated SPX data with the last 82 days and found the period coming from august 1980

Since 2 months it has perfectly shown the way

We will test it today to see if SP500 makes a 3 - 5% move up in the next week

http://groups.yahoo.com/group/futuresystem/

I am allways amazed to find charts like that and more amazed that they produce such terrific result
 
well yes it seems so simple

Now the period 1980 is quite similar than now

stocks made a bear market rallye before plunging until august 1982 (start of the long bull market)
 
Good idea...but you're might be data mining. Anyhow is has been proven that market have weekly and monthly short term memory. .. part of your theory might be linked in that respect.

Keep us posted on the result.

Regards


Quote from fortuna:

well yes it seems so simple

Now the period 1980 is quite similar than now

stocks made a bear market rallye before plunging until august 1982 (start of the long bull market)
 
I do not try to optimize I

I just select the biggest correlation (this one is 97.5% !!!) based on past data since 1955. I have other correlations (around 95%)and they produce quite the same forecasts

Memory is not only short term . Memory is a wave memory
 
Quote from fortuna:

I searched for the most auto correlated SPX data with the last 82 days and found the period coming from august 1980

Since 2 months it has perfectly shown the way

We will test it today to see if SP500 makes a 3 - 5% move up in the next week

http://groups.yahoo.com/group/futuresystem/

I am allways amazed to find charts like that and more amazed that they produce such terrific result

My data analysis, while not exactly agreeing with yours, points to similar indicators. There are several similarities in recent data as compared with the time period you mention. The interest rate environment is a big wildcard and my analysis does not show that the US Economy is structurally as weak as it was in the early 80's .... so .... we may not see an exact replay of the late summer 80's conditions IMHO.
 
Quote from fortuna:

I searched for the most auto correlated SPX data with the last 82 days and found the period coming from august 1980

Since 2 months it has perfectly shown the way

We will test it today to see if SP500 makes a 3 - 5% move up in the next week

http://groups.yahoo.com/group/futuresystem/

I am allways amazed to find charts like that and more amazed that they produce such terrific result


Fortuna,

With due respect, I believe that you are entering the realm of spurious correlation. I would advise extreme caution if you plan to trade off of this finding.

Regards,

Thunderdog
 
Ok

let's be clear

I do not trade on statistics. I just want to know them.

You might find nice correlations between SPX and a baseball cup results

However this correlation tells something and is not unique: many other assets/commodities show simlar correlation. So even if I am not able intellectually to justify this it is somewhere an indicator in my global trading plan

At least it shows that previous rallye have pushed higher before retreating
 
The forecast has been perfect today also

SPX should climb 1 to 2% until end June before giving back its 3 days gain

not an absurd scenario : mutual funds will push index for window dressing and sell evrything on tuesday

To be followed

sometimes statistics can be very excited
 
Back
Top