Socrates Forecast Array Nonsense

This is a bombshell report about what is really going on with Armstrong's Forecast Arrays system. You will not find this information anywhere else.

All users of the Reversal System will soon and inevitably be frustrated by the lack of its performance. Typically, the experience is as follows (quote):

It NEVER gives you any information on what / when to buy or sell any instruments.  It only tells you it's in uptrend or downtrend..... which is very useless

There is a very solid reason for the complete lack of performance of the Reversal System as experienced above.

It is the simple fact that the Reversal System is a trend following system, or more specifically, it is a system that follows the breakout movement from presumed support and resistance points. In fact it is not better than a trailing stop loss order. It cannot detect a change of direction, a change of trend.

Martin Armstrong would have recognized this dilemma and needed a solution to detect market turns, the difficult part as observed.

From the simple functional perspective, his Forecast Arrays are his attempt to solve the problem of detecting the next turn that reverses the direction of the current price movement.

Before I get into more implementation details, please let me give you my own conclusion after I evaluated hundreds of his Forecast Arrays.

The Fundamental Flaws in Armstrong's Forecast Arrays

The Forecast Arrays try to project the price movement into the future, often with multiple turns along the 12 columns. The problem with this is that most of the signal, the further it goes into the future, does not really matter, especially in light of the fact that the process is extremely inaccurate. It does not matter also because soon, new Forecast Arrays generated after the current one will make these future projections obsolete.

What really DOES MATTER is this:

1. The information at the left edge of whether the direction of the current movement continues or not
2. (Perhaps) when the current price movement is expected to reverse the 1st time

These two points can actually be condensed into a single question: AT WHAT TIME IS THE CURRENT PRICE MOVEMENT EXPECTED TO REVERSE? At the left edge, this number would be zero.

And this is where Armstrong's Forecast Arrays completely fail. This fundamental requirement has actually been forgotten over the years, which proves that ARMSTRONG HIMSELF CANNOT BE THE ORIGINATOR OF THE IDEA AND IMPLEMENTATION OF THE FORECAST ARRAYS.

How do I know it has it been forgotten? 

The earliest of his Forecast Arrays up to approximately the year 2015 start with a future date, one period after the current period. More lately, the 1st date is the current period which has just ended and one period before it left from it. That means that the Forecast Arrays now have two periods in the past. You might say that the design has changed and this could be an improvement, done on purpose. But this is clearly not the case because there are cases where the Forecast Arrays fail to even detect a turn that has already occurred in those two past periods.

In other words, we are looking at some kind of wreckage that has never been checked for errors for decades.

Additionally, the handling of the left edge is not even consistent which produces jitter. I have cases where the Forecast Arrays move forward in time between one and the next, but the dates at the bottom fail to follow!

The Implementation of the Forecast Arrays

Now please let me go back to the details of Martin Armstrong's solution Fortunately, Martin Armstrong confirms himself that detecting market turns in the top row is the original design idea as follows:

In How to Use the Forecasting Arrays he defines:

... Our Forecasting Arrays are the sum of all models. ...

With this rather clear representation by means of a single Array at the top, there would not be the comprehension problems that most observers of the Forecast Arrays system have. However, according to my own statistics from evaluating hundreds of Arrays, their probability of identifying only the single first major turning point is between 20% and 25%.

This would have become painfully obvious to Martin Armstrong who lost hundreds of millions trading with them. I can imagine he would have thrown a tantrum and possibly had some serious arguments with his computer programmers.

Capitulation by Exposure of Internal Details

What happened?

The final design of the Forecast Arrays has not only one chart, it has nine charts to choose from!

The additional eight rows are in fact internal details that the system owner would be reluctant to share if they had any significant value. To share such internal details is highly unusual for a proprietary system with claims of Artificial Intelligence (AI) logic. Please let me explain this by means of a single row, the "Trading Cycle". You will not find this explanation in any Socrates manual.

The "Trading Cycle" becomes very relevant if no other cycles could be detected. So if the system was dumb and could not combine its own internal signals in the top Aggregate / Composite row, then as a user, you would first need to decide whether this condition exists and then look at the "Trading Cycle" and nothing else for your trading decisions. Because the "Trading Cycle" just says "enough is enough" and creates a signal if a trend has been going on for too long after the last high or low.

But the point is that the Aggregate row logic already makes this decision for the Socrates user in such scenarios and AGGREGATES the Trading Cycle signal with its own weight. The Aggregate signal includes the "Trading Cycle". It is partly driven by it. Why then expose it and all the other signals?

Because the system did not work. The exposure of most of its internals is an additional proof of this fact. There must have been a rift between Martin Armstrong and his developers. For a limited time, Martin Armstrong might have hoped to resolve the problem by inspecting its internals in isolation.

In the end, as we know, he capitulated and re-invented his scam by deciding to sell the broken Socrates system to his clients - with a trick.

The basic concept of making this work for HIM is a variation of a tried and tested trick that has been used by con men for thousands of years. He created complex professional-looking multiple representations that he can pick and choose from in hindsight:

And the additional exposed charts fit the bill. Depending on their weight, they are often in conflict with the aggregate (composite) top row, creating peaks at different columns (dates) so the set of arrays always has enough conflicting options to pick from. This is achieved by including rows with quasi random sources such as the "Panic Cycle" and "Trading Cycle" rows. The entertaining color effect of what could best be described as a Kaleidoscope or fireworks effect is further enhanced by vertically expanding the results to fill each entire vertical chart space so there is always at least one solid bar reaching the top, and by changing the color of each row between two colors, one for rising and one for falling.

The purpose of this scheme is being able in hindsight, after the event, to pick a chart that would have identified the actual turning point as observed in reality. To support this scheme, each row is given its own plausible description to create an illusion of relevance.

So the successful interpretation of the Forecast Arrays (in a way as demonstrated by Martin Armstrong and his shills) is only possible in hindsight. That is why his computer does NOT USE its own feature, the Forecast Arrays, for the calculation of buy and sell signals. Because nobody can write a computer program that gets its rules only AFTER execution of the program! In other words, the Forecast Arrays are nonsense because all of the above has nothing to do with trading in real time.

Please let me sum it up (with some additional features). Why is interpretation only possible in hindsight? Because of built-in ambiguities as follows:

• Multiple prominent peaks even in the major signal (top row)
  
• Cycle Inversion changes (inverts) an incorrect result to be correct (cannot be predicted)
• Arrays in a single time frame change each period creating up to 12 conflicting time window versions for a single turning point to choose from
• Multiple rows to choose from
• Arrays in multiple time frames can be combined in different ways
  

A forecaster is faced with an enormous amount of ambiguity which cannot be resolved in advance.

That is why I call  Martin Armstrong the "Hindsight Forecaster".

He has the perfect smoke and mirrors machine that is always right! In hindsight, after human interpretation, which is basically the manual elimination of the wrong results (cherry-picking). 

Before I get into complicated variants, please let me start with an additional, ridiculously simple manipulation, the

Zero Horizon Forecast with Pinpoint Accuracy (after the Fact)

Please let me describe this forecast type by means of a weather forecast. Let's say it is raining today all day, and I tell you that I, as a weather forecaster, predicted the timing of the rain with pinpoint accuracy down to the day. You ask me: "When did you predict the rain?" and I answer your question: "Today, with pinpoint accuracy".

In the following, Martin Armstrong praises his computer on his public blog site for making exactly such a forecast:

On Jul 27, 2015 in Buy the Rumor and Sell the News he writes (quote, emphasis added):

...Our model on the Dow pinpointed a high for the week of July 20 ...

A simple inspection of the source of the forecast array image reveals that the forecast was made on the same date as the date of the high: DJIND-FOR-7-20-2015.png

Apart from this ridiculous accuracy claim, consider the fact that he had a whole week to carefully select a market (from over 1,000 available to him) and a point in time where the particular market and its forecast array had a coinciding high a week ago for effective illustration which is cherry-picking or curve-fitting after the fact, because the forecast was reported a week AFTER July 20.

Cherry-picking Examples

For the most recent example of Martin Armstrong's Forecast Array cherry-picking and how he actually does it in detail, see:

Martin Armstrong cherry-picks Bad Forecast Array This example also shows the failure to detect the turn at the left edge (the stuff that really matters) as I described above, but that is only part of it.

Here is another example of cherry-picking:

VIX chart

Martin Armstrong's paid shill LateralusYellow, the dictator of The reddit Armstrong Economics Fan Forum must have had great fun posting this garbage on Nov 10 2020 to dupe his audience of fools. The incredibly complex collection of Forecast Arrays in the image attached to that post (link provided here for convenience) just illustrates complete overload with garbage - something that cannot possibly called information.

For Martin Armstrong, the charlatan, the con artist, this technical garbage overload is a big loophole that allows for the sale of $160 online training material and conferences with a price  tag of $2,750 per seat when the gullible scam victims finally try to find a way out of this mess. Which I know are just empty promises because no consistent forward looking rules have ever been provided in these seminars. The charlatan only cherry-picks in hindsight and claims the computer is always right.

These factors should make it clear that Forecast Arrays cannot be used as a prediction tool. 

To the above ambiguities I must add one more huge factor: Martin Armstrong uses the well-known hat trick of cherry-picking fitting model cases that best support his point out of more than 1,000 available markets. Here is an example how he sells his $2750 WEC conference with it:

Putin Has Won (quote):

A lot of people have asked about how the hell can our computer project these targets even near-term a year in advance? Most people have a theory and then try to find facts to support it. That is why such analysis always fails. I discovered the incredible regularity of time and I did not begin with a theory so I am still trying to understand why the timing is so accurate. Here is a chart of the Monthly Array on the Dow from last July 2021. Even back then December and February were the strongest targets as you can see in the top Aggregate of 72 separate and independent models. December was the high intraday or on a closing basis in so many markets and here we are in February. Correlating everything with Russia allows us to hone in on the timing and where. It is not my personal opinion.

When we look at the Dow Array, this hinted that we could get the opposite trend from February into March and this opens the door as I said at the WEC, a high on the ECM turning point would also warn of a potential decline into 2023. Take note that May is a Monthly Panic Cycle. Clearly, a March high would not look good going forward.

There are of course inexperienced or biased critics who like to point out that I do not understand the system which would need a deeper understanding of its rules for using it successfully. They typically hide the fact that there are multiple competing conflicting rules. To those I can answer with confidence that I know the limitations of the Socrates system better than Martin Armstrong himself. To dive into more details here would be exactly what a charlatan would want me to do, which is getting into endless discussions about the system and that would require what he needs to survive: More sales of his snake oil such as training seminars.

Nobody has ever succeeded in creating consistent profits with it. Martin Armstrong would be a billionaire just trading with it. He would be owning 80% of the stock market if the system performed as advertised.

Martin Armstrong predicts market turns based on Forecast Arrays. He should know best how to use them. But as shown numerous times on this site such as in Major failed Predictions, most of his predictions fail.

Summarizing, the collection of Forecast Arrays is not a trading system, and it cannot be used to build one. It is not even a system, subsystem or model. Because every model has defined inputs and outputs, typically a single output. The Forecast Arrays combined with the Reversal System have multiple outputs, and which one to use and how to use them is subject to manual cherry-picking in hindsight. In technical terms, Socrates cannot be back-tested (for details see Backtesting) because its behavior is indeterminate.

See also:

Armstrong failed to predict crude oil PANIC