Time for a thought problem: I’ve been ponder the nature of predictive systems a good bit lately because it occurs to me that if indeed, as some have claimed, only a very few (and maybe only one) can use a predictive system at a time, then much of modern economics should have fallen by now.
I’ll give you an example: Suppose I were to present you this morning with a system of predicting market performance based on predicting future events…and further suppose that I had run some tests and it seemed to work.
Being an enterprising fellow, one might package up this way of looking at how a particular stock price could/should evolve and take it to market.
The question is: Is there a point where the very act of prediction lessens the value of a predictive system? (Hint: the answer is yes.)
I mention this because in my pursuit of perfect trading algorithms, I’ve looked innumerable ways to predict the future.
Historically many trading systems have shown keen insight into “the future” such as Elliott Wave analysis, or momentum trading, and the algorithms that keep making money by front-running trading by a millisecond or two in high frequency trading systems.
Let’s define this as our “universe” for our thought-work this morning.
Let’s suppose that a new party arrives on scene and introduces a new way of trading. We’ll call this system “Challenger” and its working are kept totally secret with the developer never revealing specific code, never sharing so much as a single screenshot, and holding tightly to the specifics of how this system works beyond flowery descriptions.
Where there is only one user of such a new trading system, it may appear to work for a some period of time. However, if the system works well enough, it will over time average down its results because more and more people will adopt it and begin trading accordingly.
Carefully note that the introduction of this new system does not end the existence of the other systems….it merely changes their behavior slightly in the short-term.
As more people adopt “Challenger” the results of other systems may even rise as followers leave them: the future is disinclined to change the total number of “winners.”
Statistically, among all predictive trading approaches, there seems to be a bounded number of winners. Only the distribution among systems changes. new systems may win early, but as traders adopt the most successful models they can find, their results suffer because of the “bounded winners” limit implicit in markets.
Let’s take this a step further: Soppose the creator of such a “new system” claims it is the only true trading system. Can such a claim be valid?
Especially because of a) the “bounded winners” limit and the pre-existence of other trading systems and b) the other pre-existing tecniques?
Obviously, multiple approaches may be used to predict the future and based on their successes, they rise or fall over time on their own merit.
In his book Antifragile: Things That Gain from Disorder, Nassim Taleb begins with a discussion of this sort of duality as he points our air may feed fire, or too much air – wind (for a candle) actually ends fire.
What he doesn’t get into the underlying principle of air. It’s the underlying variable, however.
In the analysis of trading systems, it becomes apparent (after some work) that the future does indeed seem to have a boundary to it: Although, it’s more obvious in dealing with stock market predictions than perhaps a more generalized study of future, perhaps, but a bounded future nevertheless. It suggests the future generally has these bounds.
If a new system’s results begin strong, but declaine in accuracy over time, traders will abandon it as the investor’s constant quest to optimize returns drives exploration.
That a system of divining future events claims exclusivity when dealing with something as widespread as “the future” is absurd. As they say in Las Vegas “everyone’s got a system.”
—
There is not a class of investors who stand head and shoulders above all called “Remote Viewers” at least which I’ve found yet. This hints to me that remote viewing may have limited application to investments.
Yet, can you think of any more demanding metric than returns?
Thus, the test of any future prediction system is the accuracy of the numbers it delivers.
Whatshould the sincere researcher do when faced with a future predicting system that steadfastly refuses to make tradable predictions?
One wonders which of three reasons exist: Sincere moral objection? Inability to be specific? Or, that numeric results and standard deviations really matter?
—
Lots is being written about “future seeing” and this morning’s post by Statibartfast over at Sidewalk Catharsis is closely related.
Let me be very clear on something here:
Since our Nostracodeus software has not been sold, is not yet for sale, is not deployed, etc., any failing of other predictive systems is self-arising and not related to Nostracodeus research. Any assertion otherwise is crap.
There is no blaming Nostracodeus for any “effect.” It was NOT out there and IS not out there.
In fact, Nostracodeus continues in pre-release hold.
On Wednesday, we’ll discuss further implications of the emerging Big Data paradigm world for subscribers to Peoplenomics.com.
Read More
