I think you’re probably not such a random potato as you claim, but quite a well cultivated one. This is all based on good, sensible principles and could be made to work.
I can make two suggestions.
Firstly, in my experience (and according to what I’ve heard from colleagues, too) RSI-divergence-based entry-methods for forex trading work much better and much more reliably if you make the lookback period about 22-25 periods, rather than the 14 periods you’re looking at now.
Secondly, although it’s a much smaller point, you’ll need to be very careful in avoiding anomalies if you calculate the ATR over only 5 periods. ATR-10 or ATR-15 makes only a little difference, but it does help to minimize the impact of any unexpected, individual, extra-long candles, by “diluting” their effect on the calculation.
I think you have the stochastic settings about right - something like 6/3/3 or 6/3/2 or 5/3/3 or 5/3/2 or 5/2/2 is usually about right, for this kind of idea.
I think you’re right to make the targets and stops about 2*ATR, and aim for a win rate above 50%.
Positioning the initial stop manually according the level of the most recent S/R would clearly work out better, overall, but it’s more work and much less convenient for automation and/or one-click trade-entry.