Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : Gorilla and King Portfolio Candidates -- Ignore unavailable to you. Want to Upgrade?

To: Rick who wrote (32739)	10/2/2000 9:58:28 PM
From: Mike Buckley	Read Replies (1) \| Respond to of 54805

If I throw out all the data that doesn't work with the "good" strategy, that's "data chopping." But if I, instead, just say "this might work" I simply have a hypothesis that needs testing. Exactly. There is (or at least there was) a documented, real-time historical record of the thought process that went into Ann Coleman's development of what became known as the Foolish Four that Hulbert wrote about. The hypothesis was that if Michael O'Higgins was accurate in his fundamental explanation of why the stock with the penultimate ranking was most likely to be the best performing stock, buying twice as much of that stock than any of the other three might produce superior returns than buying equal dollar amounts of all four stocks. The hypothesis was tested using about a 20-year period of historical data. It was later tested using about a 30-year period of historical data. It was simply a matter of arriving at one hypothesis and testing it. It was not a matter of data mining thousands, or hundreds, or even tens of tests and throwing out the tests that "failed." One of the really remarkable facts is that almost all of the ten or so strategies that were tested outperformed the DJIA not just for decades but also for most of the rolling five- and ten-year periods. The fact is that there was little data involved with the strategies that didn't work. If Hulbert understood the underlying fundamentals of O'Higgins's work he wouldn't be surprised to see the results of the data. I wonder how many people who have an opinion about whether or not Ann Coleman's work should be called data mining have researched and massaged the spreadsheets that put various beating-the-dow strategies into historical perspective. And I wonder how many of them have a thorough understanding of the process that was used. The article written by Mark Hulbert leads me to think he hasn't looked at the spreadsheets and doesn't know the process used to create them. --Mike Buckley

To: Rick who wrote (32739)	10/2/2000 10:38:22 PM
From: tekboy	Read Replies (2) \| Respond to of 54805

unconnected failed strategies shouldn't matter in the least depends on how you are devising your strategies. The Copernicus and Einstein analogies only hold if your strategies emerge from real theories--that is, abstract, internally coherent, deductive systems of propositions which imply the existence of certain patterns in "real-world" data. The gorilla game is such a theory: New technologies are assumed to be adopted in a particular manner, leading to hypergrowth in particular areas at particular times. Certain companies are assumed to have commanding positions in those areas, leading them to capture outsize shares of the hypergrowth's returns. The market is presumed to be inefficient in the short term (the companies' current stock prices do not reflect the likely growth) but efficient in the medium term (regular upside quarterly surprises are met with regular upward movements of the stock price). The strategy based on this theory is simple: locate and invest in gorillas and you will be rewarded with market-beating returns over time as the market gradually and incrementally recognizes what you did early on. You are correct that the records of other potential strategies for tech investing, based on other theories, have little to say about whether this one will work or not. But--and this is what I think Hulbert was getting at--many so called investing "strategies" derived from back-testing are not grounded in theories at all. "Buy the 26th-cheapest Dow stock in alternate years ending in 6" is not a strategy emerging from a theory, but rather a strategy based on hope that a particular historical pattern will persist. (OK, that's a cheap shot, but I'm trying to illustrate a real tendency.) The only reason to think the pattern might recur is wonder that it can be found in the first place: what are the odds against it? surely it can't be random! Well, if you crunch enough numbers, Hulbert is saying, eventually something that looks like a pattern will emerge, and it will only be chance. He is adding that this will be more obvious to you if I am honest at the outset and say, "I tried five thousand variations and only one worked." One other thing: any strategy that is essentially mechanical and relatively easy to follow will almost certainly regress to the mean over time as more money flows into it, trying to pick up the "sure thing." The gorilla game has suffered this to some degree, but will almost certainly not be made entirely obsolete because it requires a good amount of subjective input to operationalize on a real-time basis and because it requires a high, and unusual, degree of LTB&H discipline. tekboy/Ares@bracingforMerlin'sattack.com