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.

Politics : Dutch Central Bank Sale Announcement Imminent? -- Ignore unavailable to you. Want to Upgrade?

To: IngotWeTrust who wrote (19803)	12/15/2003 4:42:54 PM
From: sea_urchin	Read Replies (1) \| Respond to of 81118

Tutor > More along the lines of something we can agree on: If what you have written (in blue) is what you are quoting from elsewhere, then I'm not sure what we are agreeing about? Thank you for the compliment about my regression analysis but I'm surprised to hear it is "legendary" because I talk about it so seldom. Perhaps one reason is that my "models" are pronouncing that the gold bull is all but over and, in the circumstances of such widespread optimism for the metal and pessimism about the USD, I feel it is better to keep my trap shut. In fact, I believe it's more important to test the validity of the analysis than to attempt to make a big name on the internet with "I told you so" pronouncements, and then end up with egg on my face when I'm wrong! Apropos the regression analysis which you mention, I have to say that one can apply a simple regression against time to any trend and get a good fit, as the this person has done. However, the shorter the period chosen, the less information one is able to glean from the analysis unless, of course, one one is a very short-term trader. In the circumstances, I'm not sure what the person who did the analysis hoped to gain from it. Since I am, by nature and by experience, a long-term investor, I make my analyses as long-term as possible and as slow as possible. I also use a multiple regression analysis using up to 9 currency and other independent variables. I use monthly data (in fact, 5-weekly) but for a period of 25 years. This enables me to derive a kind of "overbought-oversold" pattern which I then "tune" against what really happened. In other words, when the model says "overbought", that must coincide with an actual market top (or damn near to it) and similarly for bottoms. These overbought/oversold parameters will also have a statistical probability in terms of deviation from the mean similar to Bollinger Bands or, in fact, similar to the Black-Scholes equation used for option pricing. However, I go further than statistics before I use the model to make a decision. By this I mean that if the particular analysis does not accurately define the market tops and bottoms at least 80% of the time over the previous 25 years, I reject it and try something else. I do not try to get the market to conform to the model, as most statisticians do, but I try to find the particular analysis which conforms to the market the best. In other words, before I use any analysis to help me in a decision to buy or sell it has to be "battle-tested" and shown to be right most of the time. There's also a lot of fuzzy-logic decision-making stuff which I put in to make the decisions "mechanical" and to take away the subjectivity. That's why I speak of a "model" rather than just an analysis because, in fact, it is intended to be a complete representation of the market --- all the bull markets and all the bear markets. As you can see, that's quite a tall order. I came upon this method about ten years ago and have used it successfully quite a few times since then. It accurately put me into the present bull market in gold shares in Feb 2001 and I went out when it told me to disinvest because of the strong rand. Of course, I am its biggest sceptic as I have been "burned" too many times before with other "predictive" methods which I have invented over the 40 years I have played on the market. Nevertheless, there are times when the model is uncannily correct. The problem is always whether to act on what the model says because it tells one to buy when the market is awful and to sell when the market looks wonderful. In fact, every time I have phoned the broker over the past few years to invest or disinvest, he has called me a contrarian. It takes a lot of self-training to be able to make an unpopular and "lonely" decision. It's always easier to have the comfort of the crowd.