Experimental Stock Pricing Model
August 23, 1996
Commentary. This is the first installment of the NightScope model. The underlying fundamental data and prices used to fit the model are from June 30, 1996. Ordinarily, the model will be updated more frequently, but this set of data might be interesting in light of the mini-crash in tech stocks during July.
Premiums and Discounts The model fitting process makes average premium or discount very close to 0%. An important characteristic of the model is the distribution of these premiums or discounts (technically the model views these as errors or components of price it can't explain with the fundamental variables.)
| Company Percentiles | Relative Error |
|---|
| 10% Lower | -50.93% |
| 25% Lower | -29.94% |
| 50% Lower | 1.27% |
| 75% Lower | 42.91% |
| 90% Lower | 107.91% |
By comparing a company's premium or discount (as determined by the model) to this table, one can find out how many stocks in the database had comparable, or higher, levels of premium. For example, if XYZ has 108% premium we know that it is priced higher, per unit of fundamental value, than 90% of the companies in the database. Note: For the statisticians out there, the model explains about 60% of the variance in premiums and discounts.
Please note that the model is very much in an "alpha" stage -- I am still working on model analysis and rewriting some of model building code. Look for a refit of the model on end-of-month July data and a graphical description of the dynamics between fundamentals and price.
Disclaimer
This is an experimental model. As such, the prices available on this site are only my opinion and are not official in any way. There are no warranties, implied or otherwise, as to the quality, suitability, or accuracy, of these results. This is not a quote service and I am not an investment advisor, counselor, or mentor.
Use this information at your own risk.