This is an excellent question, and I completely understand the concern, considering the history..
We manage risk in the following ways:
First is position sizing. We typically don't utilize more than 50-60% of the model portfolio. The main trade (based on the algo developed by @Romuald) is around 30% ($3k) and there are 1-3 extra trades based on Monte Carlo simulation tool.
Second is managing the negative gamma. The trades are typically opened 5-7 weeks to expiration and closed 2-3 weeks to expiration, to reduce the negative gamma.
That said, sometime even the best intentions go wrong. Last week we closed the long term bullish TLT trade that was opened months ago and went against us right from the beginning. We held it to give the thesis to develop, but unfortunately it didn't happen, and the trade was closed for $2,700 loss. Along with two other trades, we will record ~18% in October. This is probably as bad as it gets, but considering the fact that we make around 5-7% a month on average, we consider it an acceptable drawdown. We will be still up 56% in 2025 in 10 months, far exceeding our profit target of 30-40%.
Hope this answers the question.