[Project 88 Part 13] Predicting prices with supercomputers: CONFIRMED

[Project 88 Part 13] Predicting prices with supercomputers: CONFIRMED

(You can catch up here with Part 1)

At long last, our project is nearly complete.

My team of programmers has ironed out the bugs, and for the first time, we’ve been able to prove that stock prices can be predicted using pattern recognition.

Here are some important insights I can share with you about predicting stock prices:

Before I begin, I want to say three things:

1) That it’s absolutely amazing that future prices can be forecast using past prices.

That’s a big deal.

In a world where the talking heads on TV try to forecast earnings and revenue reports using “whisper numbers,” or whether or not a drug will skirt past FDA inspection…

…I think it’s safe to say that predicting where a stock’s price will go over the next couple weeks is much more important to making you money right?

2) The insights I’m about to share with you came at the considerable expense of setting up a 2000-core supercomputer that would normally cost tens of millions of dollars to buy. Lucky for us, compute time is now a commodity, so we tackled the problem well under the budget I had in mind.

3) There’s no way I could have done this without my team. It was a big risk to retire from the advisory business to focus 100% on this effort last year, but I would like to thank my crew for believing in me – especially Ruud who time and time again has been instrumental in taking my ideas from dream to reality (including making the back test literally 1 million times faster).

I know this sounds like an Oscar speech, and the music is starting to play, so let’s dive into those supercomputer insights shall we?

Throw everything you know about price patterns out the window. Profitable patterns look nothing like Elliott Wave, or any of the hocus pocus floating around the Net.

Science goes like this: You come up with a hypothesis. You (and your team) do the thousands of hours of hard, skull-numbing work to disprove your hypothesis. If you can’t disprove it, then you might be on to something.

Problem is…very few do the hard work part…because it’s hard freakin’ work, and most would rather kick back with a cold beer and watch the game.

So they end up teaching you their unproven assumptions using little more than anecdotal evidence (which is one of the reasons 95% of traders lose money).

Short-term price patterns are much more predictive than longer-term. In other words, you want a recent snapshot of stock prices just as you would want a recent photo of your online date. The results go from popping champagne bottles to eating Top Ramen if you use old data.

Take profits! Letting winners run is a great strategy, but selling at the predicted price bulks up results like ‘roids to a bodybuilder…without the ball shriveling side effects.

Sell losers fast. I’m sure you’ve heard this one before. If the stock doesn’t head towards the predicted price right away, throw it out like last month’s milk.

Don’t cherry pick sectors to trade. Throw hundreds of stocks at the strategy from dozens of sectors, and pick the top 10, 20, etc stocks that are predicted to go up the most.

I threw the entire Russell 3000 at the strategy, and it hummed along as smooth as a Swiss timepiece assembled by Daniel Roth himself.

Well, I’m just about out of room in this post…

I’m working on a report to explain how to predict stock prices in much greater detail.

It should be ready in February if all goes well.

Many thank to those who participated in our charity drive for the people caught in the crazy storms in Texas and Florida.

You will be receiving the report free with my eternal thanks.

I’m honored that so many readers of this free newsletter are so generous.

More from me soon…

Trade smart,


Dan “Prince of Proof” Murphy