Bull-like: Here's what I do. I use a data set that includes all 30 DJIA stocks, the 100 S&P 100 stocks, and around 50-60 stocks of interest that meet one of the two following criteria: a) there are 5 continuous years of increasing yearly earnings per share, or b) there is 3 continuous years of increasing yearly earnings per share if the company has not been reporting for 5 years (I get this data from the WSJ Briefing pages, and they are also available at www.barrons.com for free)
I then down-load the 52 week high and low price, closing price, pe, eps, market cap, and volatilty from the portfolio manager at investor.msn.com and run statistics on the data set (I use SAS, but any standard statistical package would work). I then calculate the predicted price and the 95% upper and lower confidence limits (LCLM and UCLM, resp) based on the 200 company dataset. The predicted price from the regression equation is what I call the strike price, or fair price, while the 95% UCLM is what I call the target.
I end up with undervalued companies (COHU, AMAT, DEVN, many others) along with overvalued companies (CSCO, PSFT, ASND, many others)
There's more details that are involved that I'd be happy to share with you, but that's the jist of it.
Your further comments and thoughts would be appreciated. |
