Post
Topic
Board Economics
Re: Martin Armstrong Discussion
by
bikefront
on 13/05/2019, 23:00:47 UTC
Yes, the gmw can be used to find opportunities, assuming they do not have a lower winrate and such. This is assuming it even works at all- it seems more backward looking and a rather lagging indicator that changes often- but again, it needs to go by Reversals for signals.

If the GMW can't be used as some type of a signal as part of a trading algorithm then I'm not sure what value it would have at all. The only way to test that is to gather the data and find its potential as a forward looking indicator which is all I am doing. Ideally, it should be tested as both an indicator for entering and exiting positions, as well as not taking any position at all. So if phase transitions and waterfall statuses are precursor for large moves up and large moves down then I should be able to quantify what it means to fulfill the parameters of those patterns and determine the probability of Socrates correctly predicting that pattern being fulfilled. I may need to look into the arrays and reversals so that I can avoid the scenarios where it may not be fulfilled which is fine. However, if the status doesn't do any better at predicting the future movements of an asset then it's just words and colors on a web page so I can say it feels like X is going to go up or down now.

And by the way, the one who wrote the imgur user guide was me. I only pretended it was not me because I wanted to have it viewed and critiqued objectively instead of being referenced. I will probably delete it soon because it simply does not work.

I'm not sure what user guide you are referring to. The user guide I was quoting is directly from the ask-socrates website.

The 30/70 rule you are referring to is in the user manual, as well:

Quote
The Panic Cycle Models identify potential timing of abrupt, possibly dramatic price movement. A Panic Cycle differs from a turning point or a directional change insofar as it does not necessarily reflect a high or low, nor is it attempting to reflect the beginning of a change in trend.

We’ve observed that approximately 70% of the time, a Panic Cycle has been an ‘outside reversal’ (e.g. when price exceeds the previous session high while also penetrating its low), or capitulation, whereas approximately 30% of the time it has been a relatively fast one way move.

(Note: the use of the term ‘reversal’ in the phrase ‘outside reversal’ has absolutely no relation to the term "Reversals" from the proprietary Reversal System available in Socrates – it simply refers to a dramatic price move in which a market price exceeds the previous session high while also penetrating its low).

For the GMW, you'll need to have an objective way of choosing your criteria, as it changes the message it displays often, sometimes daily even when it is on the monthly level. That's why I was so cautionary on it. You can look up other people having the same problem on https://www.reddit.com/r/aec/

If the Panic Cycle is either an outside reversal or a large move, and it does one or the other 100% of the time (30+70), then that means if you wait until it breaks out from the previous session to take the opposite side as well as opening the opposite trade in the form of a strangle, then it should make money 100% of the time, as the EM (expected move) should be greater than the premium value of what the market prices in. But, it doesn't work. Actually, the Panic Cycle should in theory be the most consistent array, as it does not depend on direction to make money.

Also, StrikeEagle is Armstrong. I've posted proof, unless that is some kind of ridiculous coincidence. Once, okay, but when it happens multiple times with such specificity, something is up. Armstrong should post his trades to prove the naysayers wrong. (Posting only a 90's datasheet doesn't count) Why is it that the majority are not making massive and consistent gains with Socrates?