3 Tips from 20 years of building trading strategies [Part 1]

Time sure flies doesn't it?

20 years ago — almost to the day — I built my first mechanical trading strategy.

I designed it from scratch using an old programming language called PERL.

It was two strategies in one. Mean reversion and trend following on the S&P 500.

With the hindsight of 20 years of experience, it was no where near the power of the ‘DB Transaction' I discovered 16 years later — but it was a start…

Tip #1 Let the tools do the work

Before that, I was committing the cardinal sin of not doing a computerized back test with my ideas.

That's like driving a car without oil in the engine. Sure you might get closer to your destination, but eventually, the car will catch on fire and burn to the ground.

I would do the ‘ol “back test” by hand trap, which ended up making me cherry pick trades, deluding myself like a damn fool.

A fool and his money were quickly parted in my case.

Good thing I didn't have much back then or it would have been flushed down the memory hole.

You ever do that?

Almost everyone does it in the beginning, so don't feel bad…but don't feel good about it either.

Tip # 2 Simplicity


When it comes to designing strategies, simple wins.

Take Project 88 for instance (the quest to forecast stocks with a specially designed 2000-core supercomputer).

It's pretty simple on the surface.

You look at a stock's current price patterns, and then crawl through every stock's price patterns until you have a bunch of matches…like combing through a massive FBI fingerprint database.

Then you see what happens next on average after the pattern forms.

Then you buy the stocks predicted to go up the most.

There you go — only three sentences to describe what we're doing.

But of course you need to process trillions of computations to do those things…

…and you need to back test every single day for 30 years (7560 days multiplied by trillions of computations is not a small number of calculations right)?

Digging for gold is simple in theory — just get yourself a pick and shovel and a good plot of land — but it's back breaking work in execution right?

Tip #3 Be thorough

Years ago, I read an ad about a one-legged golfer who could stripe the ball down the fairway 300+ yards like clockwork.

I wanted to know who wrote the ad, and after some digging, I found an interview of the author talking about how he came up with the concept of the one-legged golfer.

Turns out the author spent hours asking question after question from the guy who he wrote the ad for.

He kept digging and digging and digging.

About three hours into the conversation, the golf instructor casually mentioned that he learned how to consistently hit fairways from a one-legged golfer back in his early days.

Bingo! That's the kinda of information that makes you stop in your tracks, and get your ad read.

The result was a wildly successful golf educational business.

By being thorough, and poking and prodding down every possible avenue, he found the hook that would get massive attention.

When you design a trading strategy, you need to be thorough, no matter how ridiculous it seems.

And let's face it…trading is a business and should be taken very seriously when you're doing your research.

After all, you worked hard for that money you saved in a professional way, often investing several years of your life in higher education…it doesn't make sense to treat that same money as a hobby.

Often times, the way the markets actually work, and how you think they work will be 180 degrees out of phase.

Case in the point, the Smart Money Indicator for bonds.

We used the same rules as the SMI for stocks, and the results were worse than the benchmark.

Then we flipped the rules upside down, and a winning bond trading model was born.

BTW, both SMI's for stocks and bonds, and short-term bonds remain on a sell signal, putting the strategy in cash.

This post will be continued in Part 2 next week. Stay tuned…

Trade Smart,

Dan