Jim had posted a table of binned peak P/B and smoothed forward returns for BRK.
I've done some looking but can't find it, would appreciate it if someone could point me towards the latest table?
Thanks!

Bear in mind that it's only a table of PAST returns from various P/B levels.

After a few years of thought (I'm a bit slow), I decided there was a problem with this. Not only does it assume that future P/B levels will be similar to past (an obvious assumption and not too scary), and that book will rise at a rate similar to the past (again pretty obvious, and we can add a bit of conservatism easily), but this: the past tended to include quite a lot of periods that the valuation went from overvalued, promptly down through fair valued, and down into undervalued. And vice versa. So the "predictions" from that table are too extreme: they implicitly include a pretty good chance not just of mean reversion, but also of *overshoot*. I don't think that's a reasonable assumption for baseline planning purposes.

So, a much simpler method, which I think might give you a better idea of what to expect. So you don't need any old table : )

Pick a valuation yardstick. Let's say you use book per share, but it doesn't have to be.
Estimate what a normal market price multiple of that metric is. Let's say 1.4, the 20 year average, for our example.
Estimate the degree of over/undervalued today, using that metric. Let's say today's price seems to be 4% higher than the normal valuation level you picked.
Estimate what a year's growth of that metric might be. Let's say you pencil in inflation + 7%.

Those simple assumptions are enough to give you a "middle of the road" guess of the likely one year return.

Future price based on overvaluation today, drop to your "normal" multiple = 1/1.04
Factor in a year of growth, multiply by 1.07
Result is 1.07/1.04 = a one year prediction of inflation + 2.9%.

Want a guess of the most likely price two years out? Just use the 1.07 figure twice. 1.07*1.07/1.04 = up from here by inflation + 10.1%.

If you like you can do this with nominal prices rather than inflation protected ones. You might for example assume that the nominal price will rise (say) 9.5% or 10% in a given year, so just put in 1.10 instead of 1.07.

Jim