Unless OP is doing covered writes. Then OP's risk adjusted returns should be somewhat better than the underlying (e.g. BXM, PUT). However, I don't think OP is going to get any extra absolute returns compare to the underlying. If OP writes naked, you are right. And according to @Meverick74, if OP wants to write, the best is ATM because if the bet goes against OP, gamma softens the fall.
I assume one will tend to short options to capitalize on time decay and decreasing volatility as well. Why not trade strategies such as butterfly or (reverse) iron condor where net debit = max loss; the premium will increase (to a redemption value) as time goes by - and volatility goes down, if the underlying is in-the-range. "inline" warrants works this way.
If you really have to short, always go for the straddle/narrow strangle where gamma is highest and you are paid for the risk you take, never for the wings. If things go bad (and they will) at least you'll have more time to see the freight train coming.
While I appreciate what you're saying, , don't try to work an expectancy calculation on that -- it won't go well. (Hence, working the wings: working the "edge" of smallness/adroitness to take advantageous positions before the market knows them to be advantageous. )
@tommcginnis haha you got me.. Been trying to do that for a while now, and I am starting to see some promising results. So far, the best (naked) premium selling strategy seems to be a sort of asymmetric straddle (very similar to destriero's pitchforks), with the P/L curve adjusted so that it mimics the most likely distribution. Again, I play on an index that got upside/downside capped (and historically never move more than 5% in a day) so probably this doesn't translate well to US markets.
I no longer try to avoid having it go in the money by picking a strike below the current price. I sell near at the money strikes, that's where the most time value is. Then I adjust.