Amazing New Ways to Forecast the Future
Use them to make money. Consistently.

Martin here with some amazing new discoveries to share with you.

This photo of Larry Edelson and Mike Burnick is directly related. It brings back memories of them working together over a decade ago. But what it best symbolizes for me is the work they did starting three years before Larry passed away.

That’s when Larry embarked on a monumental mission. He wanted to build a computer model that could replicate his approach to analyzing cycles and forecasting future market moves.

Is this possible? Can anyone truly transfer the logical processes of a human brain to the software code of a computer? Back when I first began using computers for forecasting, the answer was almost invariably “NO!”

Today, for certain kinds of applications, the answer is getting closer and closer to “YES!”

I’m neither a neuroscientist nor a computer expert; you’d probably have to be a Nobel Prize winner in both fields to truly comprehend the fundamental differences between human brains and software programs.

However, I have personally witnessed some cases in which remarkable results were achieved even with computer models that, in subsequent years, were considered “laughably primitive.”

The first was when we created our Weiss Ratings Models back in the 1980s and 1990s. Our mission back then was, in some ways, similar to Larry Edelson’s and Mike Burnick’s: We wanted to replicate the approach of Irving Weiss (my father) to rating banks and insurance companies.

Rating Banks in the Great Depression

Irving, Abraham and Rubin Weiss, Circa 1929

Dad had originally developed his formulas back in the early 1930s when he was working as a broker and analyst on Wall Street.

Irving was like Sherlock Holmes, out and about gathering evidence. His older brother, Abraham, was like Malcolm Holmes, pontificating from an armchair. Their youngest brother, Rubin, was the artist, illustrating their ideas graphically.

To rate the safety of major New York banks, Irving spent most of his time hunting down some data. He’d walk over to a bank’s headquarters in downtown Manhattan and nag the managers. Or he’d befriend a shareholder. He’d manually enter critical numbers about assets, capital and liabilities on a bookkeeping spreadsheet. And he’d evaluate the bank’s financial strength with the combination of a few ratios he had learned or created himself.

My father’s approach was primitive compared to what we did years later. But it worked. Time after time, he fingered banks that were on the verge of failure. And time after time, he warned clients, family or friends to pull their savings out before the banks went under.

And there were plenty. In 1930 alone, 744 U.S. banks failed. By 1932, the total number of bank failures was about 5,000, with more than $100 billion in deposits lost or tied up. By the end of the decade, the crisis had claimed 9,000 banks in all. Unfortunately, however, Dad was only able to cover some of the largest ones.

Rating Banks and Insurers in Recent Decades

Martin and Irving Weiss, 1983

Fast forward a half century to the 1980s, and you’ll be surprised at how much changed. We still used many of my father’s general approach and ratios (plus many more). But …

Instead of hunting down financial information piecemeal, we began buying massive databases from the authorities — the Federal Deposit Insurance Corporation (FDIC) for banks, and later, the National Association of Insurance Commissioners (NAIC) for insurance companies.

Instead of just a handful of companies, we evaluated nearly 20,000. Instead of a few bits and pieces of data, we looked at billions at a time, always applying the same model evenhandedly across institutions of the same industry and over time.

And to make all this possible, instead of doing it manually, we used a mainframe computer.

Fair warning: 100-year old chart predicts HALF of your wealth will soon be wiped out. Click here to read the free report every investor must read before it’s too late …

Our approach was primitive compared with what we can do today. But again, it worked: In the 1980s and early 1990s, over 2,600 banks and S&Ls went bankrupt; and in virtually every case, our bank ratings warned the public ahead of time, typically three to six months in advance.

Our Weiss Ratings also provided early warnings of the giant insurance company failures of the early 1990s, such as Executive Life, First Capital Life, Fidelity Bankers Life and Mutual Benefit Life.

That’s why the U.S. Government Accountability Office (GAO) concluded that we beat all of our competitors — including Moody’s, Standard & Poor’s and A.M. Best — hands down. And that’s how our followers stayed safe during those turbulent times.

We did it again before the failures of major Wall Street firms during the Debt Crisis, including Bear Sterns, Lehman Brothers, Washington Mutual, and many more.

But if you think all of the above is noteworthy, wait till you see what’s possible now …

Artificial Intelligence and Neural Networks

Once more, you’ll be surprised at how much has changed. Compared to the giant mainframe dinosaurs of the 1980s, the data storage and processing speeds of today’s tiny computers are mindboggling.

Plus, they’re a lot cheaper. Heck, even a cheap smartphone today gives you more firepower and data access than a $1 million Cray computer of the old days.

You might think that’s merely a quantitative improvement — just more of the same basic stuff. True, up to a point. But this massive computing power has also opened the doors to a qualitative improvement — a major breakthrough that can help investors make a lot of money.

Here’s the key: In the past, we usually applied the same ratios and formulas to all companies. Except for tweaks we made from time to time to adjust to the most critical changes in the world around us, our computer models rarely changed very much.

For rating banks and insurance companies, that was not a problem: Accounting rules, balance sheets and P&L statements haven’t changed much either. But for financial markets, where change happens by the day, the hour and even by the minute, it is a problem.

To accurately predict the future, you cannot rely entirely on what happened or worked in the past. You must continually LEARN from the most recent and most current experience.

You have to promptly change your approach to reflect new phenomena. Like the Fed’s Quantitative Easing, which never existed before in U.S. history. Like government gridlock, which hasn’t been this bad since the 1920s. Like America’s political divisiveness, which is measurably worse today than during Reconstruction.

You need to quickly adjust your thinking to reflect recent tectonic shifts like Brexit and Trumponomics.

And you must instantly modify your parameters to incorporate the latest fluctuations in the financial markets, or, more importantly, to reflect the ever-changing pattern in those fluctuations.

That’s what Artificial Intelligence (AI) is designed to do. It learns as it goes. The computer program actually changes itself. And to make it possible for the program to truly find the best fit among millions of possible solutions, Neural Networks are ideally the best solution.

Computer using neural networks to survey and test millions of possible scenarios by feeding data from an “input layer” through multiple hidden layers of analysis, and then spitting results out into an “output layer.”

The neural networks used in modern-day computers are different from those used by the human brain. And they’re primitive compared with what we’ll probably be using years from now. But they work.

Here are just a few amazing examples
from a wide variety of exciting fields …

Apple: While also using an overlapping technology called “machine learning,” Apple’s Siri uses artificial intelligence to help us communicate with the Internet seamlessly.

Facebook has announced it will rely on a similar kind of logic to identify and take down fake news.

Google’s Creative Lab has released “A.I. Duet,” an experimental program that lets you play a music duet with the computer.

HiBot, a U.S. company, is combining artificial intelligence and robots to advance the way municipalities replace aging water pipes, expecting to help solve a $1 trillion problem that now faces U.S. water utilities.

IBM has trained its supercomputer Watson to effectively diagnose cancer, using massive amounts of data from multiple medical research sources. A few miles away from our own offices here in Palm Beach Gardens, the Jupiter Medical Center has just signed up for the service.

Or consider these applications that may
be even closer to what you do or want to do …

Credit card fraud detection: You use your credit card to make a purchase. Maybe you’re overseas, or just online. Perhaps you’ve made several purchases within a short time frame and/or you’ve entered your password incorrectly on the first try.

Result: Your transaction is blocked. “How did that happen?” you wonder. The answer is that some kind of program, possibly using neural-net software (albeit still quite primitive), tried to put all the bits and pieces together, picked up something that’s different from your typical behavior pattern, and raised a red flag.

Foreign language translation: You want to translate a phrase from, say, English to Spanish. Not long ago, even if you used one of the world’s most advanced translation programs, such as Google Translate, the errors were often laughable.

Today, that’s changing rapidly, thanks in part to neural networks. Last year, in fact, Google announced that it was using Neural Machine Translation (NMT) to translate more complex sentences. They say the improvements helped reduce translation errors by 55 to 85 percent. And that’s just the beginning.

Accounting: In 2015, the World Economic Forum surveyed 800 executives and found that three out of four believed that, by 2025, 30% of corporate audits would be done by intelligent machines. Of course, humans would do the rest of the work. But 30% is a huge savings in manpower. And today, that same survey would probably produce results which are even more promising.

Stock ratings: Every publicly-traded company must submit reams of financial data to the stock exchanges, the SEC and the IRS. To understand them in context, that data then needs to be compared to similar reams of data available on dozens — sometimes hundreds — of companies in the same industry sector.

This, in turn, means little without years of history. And that doesn’t even begin to consider the trading pattern of the shares themselves.

HALF the risk, DOUBLE the profit potential! Paying 50% less for an investment position means you cut your risk by half. It also means you DOUBLE your profit potential. No wonder so many wealthy investors say this is the ONLY intelligent way to invest! Read more here …

With artificial intelligence models, all of this is NOW possible almost instantly. I know. Because that’s what WeissRatings.com does every day. Most important, with the help of neural networks, the results can be directly translated into highly profitable trading strategies.

Global forecasting: Investors have poured money into Boston-based start-up Kensho Technologies. Why? Because Kensho says its artificial intelligence software can comb through huge amounts of data to find correlations between world events and the resulting effects on publicly traded stocks. Whether they can do so consistently or not still remains to be seen. But we do know one thing: The start-up was recently valued at more than $500 million.

Larry Edelson, 1954-2017

This is the type of effort that Larry and his team pioneered with in-depth analysis of cycles. And it’s thanks to their prescient cycle analysis that, right here at Money and Markets and Weiss Research, we were able to publish …

• Their November 1999 forecast that the 19-year bear market in gold and gold shares was ending and a major bull market was about to begin.

• Their August 2011 forecast of a major decline in gold and gold shares.

• The huge profits that gold share investors could have made between 1999 and 2011 based entirely on their forecasts — 850% on Agnico Eagle, 878% on Kinross, 1,059% on Newcrest, 1,248% on Goldcorp, 2,958% on Royal Gold, and many more.

• Their accurate forecasts of nearly every major crash and bull market since 1986.

• Their unbelievable forecast made in April 2012 that the Dow would reach 20,000 by the year 2016. (The Dow hit 19,988 in December 2016).

Hard to believe? Then see the details in “4 Bold Forecasts on the Dow, Gold, Bonds and Euro.”

Fascinating, isn’t it?

Now, brace yourself for what we can do with artificial intelligence and neural networks!

Good luck and God bless!

Martin