Feb. 6, 2021

You design an automated stock trading strategy and will use historical data for your simulations. Right off the bat, all the stock prices you will use are part of recorded history, and therefore, what kind of “discovery” are you going to make should be the question?

All the data is already there in plain sight. All you have to do is access it. Somehow, for some, it is as if the price of AAPL over the last 20 years has eluded them. As if they had never seen it before or did not know what it did or what it stood for? AAPL and its related data are there, and that is up to yesterday. Period. We can immediately see the hindsight problem this can create.

Moreover, it is not that you have not seen it, even as a newcomer to the game, that it is not there. The historical AAPL price series is the same for all to see and analyze in whichever way they want. We certainly do know that it has been going up for years. And the fact we know this will bring some bias to any trading strategy we might want to design. We cannot escape that we already know the past, especially if we deal with it every day of the week.


The first thing to resolve should be: What kind of “discovery” could you make that is not already “discovered”? Some unusual pattern that repeats once in a blue moon? Or, some other anomalies that have already been exploited by thousands and thousands of other market participants? What can you consider as your “discovery” if almost everybody else already knows about it? Furthermore, how would you know if the strategy you are presently designing has not already been “discovered”? Or, a more simple question: who determines that what you have is a “false discovery”? And this without even the knowledge of what is a “true discovery” or even just a “discovery”.

In this trading game, the future will be different from the past. And therefore, your trading strategy cannot give the same answers as your simulations. Not because you made a “false discovery” but simply because all the price data will be different going forward.

You Are ONE

You have a game you want to play, and you are on your own. There is only one of you. In the other “camp”, there are millions, all wanting to also make a profit. For “them”, your money is almost considered as their own since that is where they are going to get it from. It is just a question of time. You are expected to lose just like the other 80% or so of “them”.

You need to design something that will put the odds in your favor. Otherwise, you will become “their” lunch. And this strategy of yours better be able to survive what will be thrown at it over the long term. The prime directive should be: at no time in the next 20 years should your trading strategy ever go bankrupt for whatever reason. Evidently, if it does, it is game over and you wasted a lot of your time and money. Not because you were “unlucky” but simply because you did not plan for what could come your way. In a way, for not doing your job!


Your stock selection process includes AAPL in its list of tradable stocks for whatever reason, isn't your strategy not biased just by selecting it? And if your stock selection does not include AAPL, are you not ignoring history? Would it not be the same for any other stock you might select? So, you find some general rule that would select AAPL for your portfolio and then give that “excuse” for its selection. Again, some hindsight at work!

Avoiding false discoveries is a misnomer. You cannot do a stock simulation on past market data as if you did not know of their existence. The simple fact, for instance, that you know that stock prices have been going up over the past 10+ years is sufficient to declare your trading strategy as biased just for picking stocks. US stocks have been in a long-term uptrend for the past 240+ years. So, what would be the surprise if this went on for another 20 or 40 years? Shouldn't we expect that? Shouldn't the world survive and continue to prosper?

I've designed trading strategies for over 30 years. Intentionally giving all of them a built-in upside bias making the same bet Mr. Buffett has made a long time ago: “Do not bet against America”. Is it biased to design an upside-biased trading strategy in a secular upward trend? The long-term should have been one of the primary concerns in the initial design phase. Do you design your trading strategy for an up or down market going forward?

US Market Secular Trend

The US market secular trend has been around 10%, compounded with a historical standard deviation of about 16%. If we put this in a chart for the next 20 years, we would get:

Estimated DJIA 20 years at 10%

DJIA 20 years

The above chart shows the range the DJIA might have over the next 20 years. With a 95% confidence level (+/- 2σ), you might consider the DJIA to achieve between 119,125 to 319,336. That is quite a pretty wide range! That was not a difficult chart to produce.

But a more pertinent question might be: will this secular trend persist? Say the future grows at only a 7% rate; what then? That chart too, is simple to produce:

Estimated DJIA 20 years at 7%

DJIA 20 years 7%

We can see from the above chart that there is quite a difference over the long term, even with just a small 3% growth difference. Now, the chart goes from 68,521 to 183,683 (a 73% difference for that 3% long-term CAGR decline).

The question should be: what is your trading strategy to do in order to exceed the performance of the first chart above? Because, if your trading script cannot produce better than that chart, you again wasted your time and money since you could have simply bought a low-cost index fund and done better.

I would say use anything you can find, discovered or not, that will give you the confidence needed to apply your strategy for years to come. So, build that strategy on solid foundations and make sure it can survive the unexpected. At times, cash is also a strategy.


Related articles:

Winning The Automated Stock Trading Game

Stock Portfolio Backtesting

Feb. 6, 2021, © Guy R. Fleury. All rights reserved.