There is a common misconception with options (at least, with selling options) that you need to be as correct as possible in your trades. That is, you want to maximize your win rate–how often you’re on the right/winning side of a trade. After all, if you could perfectly predict a stock’s future price movement, then you could sell puts at the optimal strikes every week and *never* be assigned shares (never lose). This would be ideal because you’d be maximizing your weekly premium while never actually taking on shares/stock exposure.

In essence, this is what my algorithm is trying to do. It’s trying to pick the best possible options (and strikes) each week. Best = highest probability of returning the most profit. But guess what: my algorithm isn’t perfect. It’s only “right” about 70% of the time. Look at the plot below:

It shows my win rate over time. Win rate is essentially the number of puts that expired worthless (a “win”) divided by the total number of puts sold (expired puts + assigned puts). As you can see, there were some pretty significant swings in my early days–a common feature when dealing with small samples. As the number of trades increased over time, my win rate evened out; the fluctuations decreased and the win rate is now in the 70-80% range.

Meanwhile, over the same period of time (about 15 months as of this writing), my return on investment (ROI) steadily increased. The plot below shows my total ROI over time. As you can see, there were bumps and drops when I was assigned on losing stocks (about 30% of the time), but eventually those recovered (with the help of covered calls).

I’m perfectly OK being right 70% of the time. As long as I’m collecting ~1-3% in premium every single week, I predict that I will come out ahead. So far, *that* prediction is 100% accurate.