Correct me if I am wrong, but the underlying assumption is that the price is "mean-reverting" to the trendline based on a scaled version of book over the past ~20 years. The idea is that when it is very much above trend (as it is right now), then the stock is likely to trend towards the trendline price. As such, the future price should be less than the current one, which would make the in-the-money calls profitable.

Yes, it's a pure mean reversion play. The in-the-money call will be profitable, including the in-the-money portion of its premium, if the stock goes back to its "normal" zone. But, a bit more subtly, you are going to be doing this repeatedly over time. Eventually the stock will get back to its normal zone, so (a) you will make that in-the-money part of the premium sooner or later, and (b) on every round you will be pocketing the time value in the option, which adds up. The only gotcha is that the fair value of the stock is rising over time, so the "normal" zone is rising, so your target strikes will be slowly rising too. Closing a call and replacing it with a new one could lose money if the new one has a smaller premium than the old one because fair value (and strike) have risen. This is unlikely, as value rises slowly.

My question becomes one of rules of thumb, when the price is so far out of trend, it seems that the trendline + standard deviation as a target is too low, and the put price likewise very far out.
Any recommendations on what to do in this case?

One strategy would be to always use prices far from the expected mean level.
There is a trade-off between high rate of return on the time value of the option you write, and how far it is from the current price. If one chooses options with very high or very low strike prices your chances of getting those options exercised is low, but (a) the return is so low that it's likely not worth the bother and (b) what's wrong with assignment? The strategy suggested has cash to back up the puts being written, and stock to back up the calls being written, so any time an option is exercised, celebrate! you got every last cent of the time value, and you got it early. Just sell (for the exercised puts) or buy back (for the exercised calls) the stock, open a new option position, and repeat. The system does not rely on simply writing options all the time which are going to expire worthless. There will be lots of contracts that get exercised, it's part of the normal set of trades to be done.

The +/- 1 standard deviation number was chosen a bit arbitrarily, simply to get the idea that it will be in that zone a certain percentage of the time. You could use +/- 0.7 SD, or +/0 1.5 SD, or whatever you like. Nobody really knows what the variation in pricing will be in future, and value metrics are always a bit limited in precision, so you do need a range that is of a decent

As to your specific question, "expected value plus 1SD" seeming low because the current price is already above that and has been for a while...I think it's just a matter of getting your head around the idea that you know better than the market does. Berkshire has the unusual property of trading within a fairly narrow range of valuation levels, or at least that has been the case since the credit crunch, so the underlying leap of faith is that the range in the next while won't be wildly different. If you're wrong about that you don't actually lose money, you just don't do as well.

Bearing in mind that I haven't done this as a mechanical system but simply as an "off the cuff" trade from time to time, what I do is this: if I think the likely price 3 months from now is below the current price, I generally write a call at the money, since that's where you get the most time value. But if that call is replacing one that I've lost money on (I'm rolling the position up and out), I will tend to pick a strike that is sufficiently in the money that the trade does not eat cash. This isn't being in denial about losing money on a trade, I don't care about that, but done as a regular rule it means that the original block of "in the money" premium as mentioned above is always still there to be recouped, eventually, when the stock gets back to its normal valuation zone.

FWIW, the stock has been more than one standard deviation above its smoothed value estimate for about 11.7 months now, the longest stretch since the analysis period starting Jan 2007, so writing calls has not been the best idea in this stretch. (ask me how I know!) Perhaps there has been a regime change and valuations are going to over around a new higher norm and the 20 year history is not a good guide. But I have a hunch that it's simply that Berkshire has been dragged up along with an exuberant market, and it won't last any longer than the exuberance does. Berkshire's stock-specific return has its little ups and downs, but the total return since Q1 three years ago is exactly the same as for SPY.

Sorry, kind of a rambling response.

Jim