I think it's relatively safe to say that life expectancy will be higher in 20 years...Btw, health outcomes are quite relevant to future economic performance.
While I'm inclined to agree with both of these statements (especially the latter), whatever message you were trying to convey went right past my thick head.
I'm skeptical of calculations to 1/10 of a percent accuracy based upon extrapolations of the past into the future. The history of forecasting precise numbers is not reassuring.
You have a much greater mastery over the art of euphemism than I could ever hope to attain. Coming from a background in the sciences and engineering, forecasting in the social sciences strikes me as - oh, I'll try it - what you said!
But this does bring us back to an issue that is central to the enterprising investor (I hope ya'll are willing to discuss the enterprising investor as well as the conservative investor) and that is the source of the margin of error we seek which will facilitate our outperforming the conservative investor by a sufficient amount to overcome the costs of taxes, transactions, and of course, time.
Earlier I offered a post giving VL's actual record on predictions for 49 large companies. Of course we could make our own predictions or use somebody else's, but I think that at least for purposes of this discussion we can safely assume that the track record would be no better.
Here are some questions I think we should attempt to answer. I will try to play with these, purely in a mathematical sense. Perhaps others will also try some mathematical modeling, but also, how about some non-mathematical thoughts?
1) Given the expected scatter in results, what is the minimum number of companies one should hold?
2) Given the goal of earning extraordinary returns, what is the maximum number of companies one should hold?
3) What guidelines can we elucidate for the margin of error required as a function of the certainty of expected results materializing? i.e., KO and PG are very predictable; GM, CAT, and BA are not. For an even more apples to apples comparison, HWP is considered far more predictable than IBM. If this holds (and I'm disinclined to debate it, for it is irrelevant to the question) then how much greater a margin of error should we require for IBM as compared to that which we would require for HWP?
Axel
|
