Subject: Re: Value, when to buy
If I could show a third order polynomial trendline, ...

FWIW, I used to do that a lot, looking at the correlation between past returns and initial starting value and building nice fits.

Ultimately I decided that there was a different way to come up with a reasonable expectation for forward returns that was both more reliable AND vastly simpler to do.

First, the comment about more reliable: the problem with historical forward returns is that in the past, the stock has tended to overshoot. After a period of high valuation, the slide has on averaged continued right on down past the typical valuation range into the cheap range, often very cheap. Thus, the historical record shows that the average return after high valuation levels is not merely bad, but ghastly, as the end point has on average been at a point of substantial cheapness. This overshoot effect might happen again, but more to the point, it might not. A far more conservative and reasonable assumption would be that the end point is merely a normal valuation level, not super cheap.

So, the vastly simpler method. First, pick a value yardstick. I'll use book per share as an example, but it could be whatever metric you prefer. First, estimate what the average valuation multiple will be in the future dates you're interested. Let's say you think that the average P/B in the next few years will be 1.45, a bit higher than the average in the last 20 years because, well, recency bias? Doesn't matter, pick any number. Now, estimate how fast your metric is likely to grow in a typical year. Book per share, as with almost any decent value yardstick, has grown at inflation + 8%/year for ages, so let's pencil in inflation + 7%/year for a bit of conservatism. Then it's pretty easy: Maybe you want to estimate what the likely stock return will be for the next 2-3 years. Take today's level of your preferred metric, add two or three years of growth, multiply by the valuation multiple you think will be normal, and you get a likely future price. Divide that by today's price and annualize the result, and you have probably the best single estimate possible of the likely return in the time frame you're concerned about. Voila!

e.g., using the example figures above, known book per share is $455055 at the moment. Two and three years of 7% growth give future book of about $521k and $557k in today's money. At 1.45 times book that means prices around $755k and $808k in today's money. The price is $715720 at the moment, so the estimated likely 2- and 3-year returns are inflation + 2.7%/year and inflation + 4.1%/year respectively. TIPS would probably get you around inflation + 1.5% in US dollars, so Berkshire stock at this price is a better bet, but not going to make you rich. As the current market price falls, the likely forward rate of return obviously rises.

A fancier share valuation model might improve that guess a bit, but in the end I concluded that fancier methodologies won't. There is no real need to peer really deeply at the historical record other than to inform your sense of what constitutes a reasonable expectation of future valuation multiples. It's true that this method requires you to make a couple of very explicit assumptions about things that are not certain, but other methods make them too, just a bit less obviously.

Jim