I'm starting to experiment with covered call strategies, and I'm trying to find the right strike price to sell for my covered calls such that I can maximize premium while being generally "confident" that the strike price won't hit and I can just bank the premium. "Confidence" would hopefully be an adjustable parameter in my calculations; I.e., maybe this week I'm willing to sell with 70% confidence whereas next week I'm more conservative and willing to sell calls with 95% confidence that shares won't be called away. My plan is to write weekly covered calls against steady/boring unsexy stocks.
I downloaded weekly adjusted close data and calculated the weekly absolute price move and the percentage move from one week to the next for the last 2 years. That became my sample set.
Here are some things I considered:
- I started by trying to calculate the normal distribution (sigma) for all of the absolute price changes over my sample. I was thinking that if I could settle on a price that's 2-sigma higher than the current price, then I'd have 95% confidence that my shares wouldn't be called away. But this is a pretty rudimentary approach, and potentially skews more to the positive side than negative (or vice-versa) depending on recent stock movement. I don't think this is a good approach.
- So then I tried separating the positive moves and negative moves, and doing a normal distribution over those samples. I was thinking If I set my strike price at
current price + average positive move + sigmathen I'd get a 68% confidence that my underlying would stay below the strike price. I think that method skews artificially high though, just based on where the majority of the distribution occurs if graphed on a number line.
- I also considered using a percentile of price moves over that sample and basing strategy on that, i.e., just choose the positive move that occurs at the 85th percentile and set the strike at
current price + 85th percentile price, but... I don't think that's a statistically sound approach either.
So I've done some homework, but I'm pretty sure I'm not on the right track. And statistics are not my strong suit.
Is there a well-defined way to choose strike prices that meet a statistical confidence threshold like what I'm describing?