Subject: Re: Seth Klarman on CNBC
Berkshire isn't expensive. But is it ever?
I consider $359.52 intrinsic value. Peak book to date of $231.95 x last four quarters peak P/B of 1.55...

Fair question. I'd say so.

I personally define a "fair price" for any investment as the price that gets you inflation + 6.5%/year from your purchase date.
That's basically Siegel's constant, the historical average real total return from the average US stock in the average year.
You can choose different time horizons for the investment, but I prefer to pick the average price over a fairly long stretch as the end date, so it isn't dependent on any one lucky or unlucky exit price.
So, for example, if you take a purchase on a given date, and compare that to the average real price 4-10 years later, that's like a super-smoothed endpoint on a 7 year hold.
You can annualize that as a 7 year rate of return.
If that's more than inflation + 6.5%, Berkshire wasn't overvalued when you bought.

On that measure, Berkshire hasn't been overvalued very much of the time, at least not since the late 1990s bubble pricing ended, but sometimes.

There was a fairly long stretch March-August 2007 that P/peak-book values were below 1.55 but you still got lower than average inflation + 6.5%/year in the next 4-7 years.
The lowest P/peak-book in that stretch that led to a lower average return was 1.47.

Quite a few stretches in 2005-2006 as well.
The lowest P/peak-book that got you under inflation + 6.5% was 1.432.

So, based on history, and using my dubious definition, sometimes valuations below 1.55 times peak known book turn out to have been "overpaying".

Historically, since 1997:
a purchase on a day with P/peak-book under 1.55 (50% of all days) had about a 32% chance of not hitting my real return hurdle.
a purchase on a day with P/peak-book under 1.50 (42% of all days) had about a 20% chance of not hitting my real return hurdle.
a purchase on a day with P/peak-book under 1.45 (34% of all days) had about an 8% chance of not hitting my real return hurdle.

So, at least with the very modest valuation multiples we have seen in recent years, and my dubious definition of "Fair price", one could make a case that Berkshire is sometimes "overvalued" even at P/peak-book in the 1.45 to 1.50 range.

The good news:
The average ending price method I used to calculate the returns assumes you have no discretion about which day you sell: you don't consider valuation levels.
Surely in any given 6 year "endpoint" stretch there are some good valuation days which would get you a return better than selling the average day.

Jim