Bradley siderograph

[DSteckler]

Some call this nothing more than astrology but it's amazing how accurate the turn dates can be.

http://www.amanita.at/e/faq/e-bradley.htm

[IIC]

Some call this nothing more than astrology but it's amazing how accurate the turn dates can be.

http://www.amanita.at/e/faq/e-bradley.htm

+/- 7 days or less...Hmm...The stars say I can do it +/- 5 days or less Dave...Of course I'm an a.m. Aries...Doug

[DSteckler]

Usually +/- 4 days. I get the feeling you've never seen this before, Doug. Am I correct? Don't be too quick to dismiss the siderograph, it's amazing how often it is spot on. Just keep in mind that a turn date doesn't necessarily mean a reversal; it can mean an acceleration of the preceding trend.

[IIC]

Usually +/- 4 days. I get the feeling you've never seen this before, Doug. Am I correct? Don't be too quick to dismiss the siderograph, it's amazing how often it is spot on. Just keep in mind that a turn date doesn't necessarily mean a reversal; it can mean an acceleration of the preceding trend.

OK...I'll look at it again

[IIC]

OK...I'll look at it again

But I'll tell 'ya...when I see Astrikos mentioned...I have my doubts...Remember them???

[DSteckler]

<< when I see Astrikos mentioned...I have my doubts...Remember them??? >>

Sure do.

[billyjoe]

11 major turning dates in 2006 , windows +/-4 days sometimes +/-7 days. Couldn't that account for about 1/2 any time period over 365 days? Sorry, but this one is too far out.

billyjoe

[DSteckler]

Billyjoe, let's look at how past years performed according to the Bradley Siderograph.

The predictions for turn dates in 2004 were 1/19, 1/26, 2/26, 3/6, 4/26, 5/17, 6/17, 6/22, 7/30, 8/11, 9/12, 9/28, 10/24, 10/30, 12/3 and 12/18. The dates in bold were major turning dates. Please keep in mind that the dates do not predict polarity and that a trend may accelerate on a turn date, not reverse. Now let's look at what hapened to the SPX on those dates.

1/19, 1/26 - 1/19 was a Sunday. On 1/20 the SPX closed at 1139. On 1/26 it peaked and closed at at the high, at 1155. It dropped 30 points in the three days thereafter.

2/26 - not much happened
3/6 - market peaked on 3/5, dropping from 1163 to 1087 on 3/24.

4/26 - market peaked on 4/27, dropping from 1147 to 1076 on 5/12.

5/17 - Within 5 days of the bottom on 5/12, market reversed and rose from 1084 to 1142 on 6/8.

6/17, 6/22 - market topped on 6/24 and dropped from 1146 to 1079 before bouncing higher on 7/26.

7/30 - market topped on 8/2, dropping from 1109 to 1062 on 8/6.

8/11 - market bottomed on 8/13, rising from 1061 to 1125 before consolidating on 9/10

9/12, 9/28 - consolidation from 9/10 to 9/21. Bad signal on 9/12. On 9/28 the market bottomed and ran up from 1101 to 1142.

10/24, 10/30 - market bottomed on 10/25, running 41 points higher by 10/30. The move accelerated on 11/2, running up to 1197 on 12/3.

12/3 - market topped and pulled back until 12/9 where it reversed and went higher. Bad signal.

12/18 - market made a short term bottom on 12/20 at 1193 and moved up to 1217 on 12/31/04.

So there were 16 signals in 2004, 1 of which went nowhere, 3 of which were wrong, and 12 of which were correct. That's 75% correct, a pretty good showing.

[billyjoe]

Dave,
According to my calculations with a +/-7 day window you're automatically correct 61% of the time using 16 signals in a 365 day time period.

billyjoe

[DSteckler]

Billyjoe, that assumes the time interval between signals is equally spaced. That isn't the case.

[ParkTwain]

Oh God, EW is bad enough. So what is the definition of a "turn date," like something that in hindsight turns out to be important. Very scientific. These people won't go broke selling folks what they want to hear.

[Gatorman]

Oh God, EW is bad enough. So what is the definition of a "turn date," like something that in hindsight turns out to be important. Very scientific. These people won't go broke selling folks what they want to hear.

Not a dissimilar theory from Cramertocracy which involves reading the number and direction of bumps on Jim Cramer's follicular challenged noggin'.
There are, of course, a somewhat vociferous group of afficianodos who are able and willing to help add to the number of bumps on his head at any time!

[Karel]

Back on topic: the methodology suffers from the Asterology Complaint: it may not be too difficult to add a nice story to the numbers. How do you check? I looked at Dave's commentary of the predictions and saw that a 2.5% difference in a period of 7 market days could qualify as a turning point or acceleration. I then dropped the SPY closing prices from Yahoo in a spreadsheet (should have been ^GSPC perhaps, but I prefer to account for the dividends) and looked for points fitting those criteria. I then added some more "turning points" by looking at the chart (these have dates mm\dd), because I didn't want to be too critical. If these extra points were predicted, it was OK, if they weren't, we forget about them. These were the points:

01/26 -
02/04 +
03/05 -
03/18 -
03/23 +
04/12 - (04/05)
04\20 +
04/23 -
05/05 -
05/17 + (05/10)
06/03 +
06/30 -
07/20 -
07\26 +
08/02 -
08/12 +
09\17 -
09/27 +
10/06 -
10/25 +
11\02 +

And this is the Bradley Siderograph results revisited, combined with the list above:

01/19 False.
01/26 OK
(02/04 Missed)
02/26 False
03/06 OK
(03/18 Missed)
(03/23 Missed!!)
(04/12 Missed)
04/26 OK
(05/05 Missed)
05/17 OK
(06/03 Missed)
06/17 False (two signals for one event is a bit too much; and this one is the dud)
06/22 OK (market drops, rebounces and drops definitively on 6/30)
(07/20 Missed, could easily have been counted as "acceleration")
08/02 OK
08/11 OK
09/12 False
09/28 OK
10/24 OK
10/30 OK
12/03 False
12/18 OK

Summary 11 OK, 5 False en 7 Missed, for a score of under 50%

Regards,

Karel

[DSteckler]

Huh? I don't understand what you did here, Karel.

Bradley turn dates are forecast a year in advance; they're not a reaction to a market drop/rise. The 2006 dates are shown here:

http://www.amanita.at/e/faq/e-bradley.htm

[Karel]

That is alright. I just looked at 2004, looked at what were claimed as "events", looked for more events according to the criterion I developed, and noted there were several events the prediction missed.

Now with criteria that result in more events than are being predicted, the chances of hitting an event with 15 day windows are raised. And statistically, on average 60% of those hits would fall in 9 day windows.

So I am not impressed.

Regards,

Karel