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.

Pastimes : The Death of Silicon Investor -- Ignore unavailable to you. Want to Upgrade?

To: E. Charters who wrote (885)	9/29/2002 12:07:45 PM
From: Toby Zidle	Read Replies (1) \| Respond to of 1003

Tomes's methodology sounds very promising. Is there a website that gives more detail or, better yet, charts various stocks with the application of his methods? Linear regression analysis is available through a number of software packages. It may even may on charting web site, though I can't name any. It's available on my software though I spend practically no time on it at all. I'm just oriented toward other tools I find useful. Linear regression analysis, as I know it, is far different from Fourier Transform Analysis. FTA gives an output that is a frequency amplitude spectrum and a frequency phase spectrum. From the frequency amplitude spectrum, you can determine the dominant frequencies of the preceding price chart swings. If you have no reason to doubt that they will continue in the future [a cash takeover offer would be one reason to doubt], you want to generate a sine wave of the dominant frequency and splice it into the front end of the time chart with the "least error" phase. That gives you a statistical 'best fit' curve of future price movement. If you have this, then you know when to expect (with some statistical error) the next high to sell/short and the next low to buy/cover. Then you don't sell in panic three days before the downtrend turns up and you don't hang on in greed when your stock can't break through resistance. There are other variations to this theme. Fast Fourier Transforms (FFTs) work on a number of samples that is any 'x' power of 2. If you have a history of 256 points (just over a year's trading), you get a clean FFT. You have well-defined amplitude and phase spectra. You then chop off the oldest data point and have the computer find you a new price point (tomorrow's close) that will be the closest in reproducing the previously determined amplitude/phase spectra. Iterate this 22 times and you have a decent moving price target for the next month that shows you the dates of the expected highs and lows in the trading cycle. With enough computer power, you can compute your 256-point FFT ending each day in the past month. Then you can see how the FFT is changing dynamically. You model future FFTs to continue the dynamic change. Then you compute future prices to model a dynamically changing FFT instead of the static one I was using in the paragraph above. I think you can get the picture. Anyway, I don't know of any software or any web site that does this. Maybe Tomes comes close. I'll have to check him out. What's his URL? Now to the lurkers here -- If any new software package or web site suddenly comes out that offers Fourier Analysis projections, you got the idea here first! I want to discuss royalties - or at least a program credit. LOL!