[Project 88 Part 11] Deadly assumptions

Deadly assumptions

It was Monday October 19th, 1987.

Stock markets start falling in Hong Kong, then Europe, then the United States.

Suddenly the Dow went from down a few percentage points to down a whopping 22% in one day.

There were at least two suicides in the following days.

The news media blames a wide variety of reasons: The elimination of tax benefits, Iranian missiles, a closed London exchange due to a storm.

But they're all wrong.

The reason for the crash was a simple assumption about how the market moves.

A math error.

When quants first began trading index futures in the early 80's, they made a horrible mistake.

Inside their risk equations, they used statistical analysis that only works on normally distributed data.

You know — a bell curve.

So the expected odds of a 1987-style crash was 20-sigma (10^-89)… the equivalent of picking a single atom in 1 billion universes.

All the MBA's were taught the same sh^tty assumption by their ivory tower dwelling professors, and were over-leveraged when an above normal correction came.

Selling begat forced selling.

It was clear as day looking back at the COT futures data.

The key takeaway is to be careful of your assumptions.

In that regard, I've been testing for the most significant price patterns (which I'll be sharing soon).

We had to make sure to use extremely clean data. We're dealing with over 50,000 instruments, and many are thinly traded, which wreaks havoc on tests.

Next, you have to go through the data, and find the patterns that make the most money…

And then — and this is important — make sure that the odds of that pattern happening randomly are very small.

In science, you have to setup a null hypothesis, which is theopposite of your hypothesis.

My null hypothesis is that the top pattern found is no different from picking numbers at random from our 50,000 stock database (if correct, then Project 88 is DOA…yikes).

Since stock returns are NOT normal, you have to do some statistical testing tricks, and convert the numbers into something manageable, like this:

Top Pattern #1 Magnitudes after a pattern match is found

Turns out that the odds for the top pattern in test #1 being due to randomness are 3.79*10^-37.

You have better odds of winning the Powerball Lottery three times in a row.

So why the heck am I even telling you about this anyway?

I've already proven that Elliott Wave is a sham right?

No better than random.

Disproving Elliott Wave and Fibonacci retracement waves is one thing.

Proving there are significant, non-random patterns is another.

This is the first extremely solid PROOF that Project 88 is finding off the charts significant patterns, and I wanted to share this moment with you.

I'm not popping champagne corks just yet…there's more hard work to be done, but our to-do list is sure getting shorter and shorter.

More from me soon!

Trade smart,