Bhai !!! first you said entry will be by discretionary and exit or tsl on mechanical way, this is complex way IMO,Mechanical is always simple,main issue is with the DD,which trader has to accept if he wanna follow the method. any trader including me started with floor pivots,support resistance,chart patterns,price action with candle commentary as well as structural pivots (whether vp's and mp's or normal swing points), i settled with Market Profile along with Demand and Supply !!!
Yes !!! tight stops don't work, ATR or donchian channel is good idea to some extent,Many are using tight stops with Structural pivots , (SL below or above MP) ,lot depends on back test and traders mind to accept the thing !!!
Yes discretion in entry selection adds some complexity over simple mechanical signals, but it can also add some value in pattern recognition and context.
Drawdown is for everyone, i dont think discretion can eliminate it, maybe mech has more but in turn it is probably less emotional in day-to-day trading, esp if automated.
Yes, Discretion can also add a lot of confusion. I had very tough time with NiftyNirvana and also with Pivots + tight stops. Pivots is more structural and less PA. Made progress only after following Adam Grimes for a few years - really great book and blog among other things.
Just saying that mixing discretion + mechanical can also work - there is no problem there.
Even for entry - mechanical helped me understand how my market and TF works and then i use that as info for my discretion.
Ex Open hour in stocks is mean reverting, or atleast not trending for my setups/TF. I use that to avoid trades in open hour and then look for my kind of pullbacks later.
But the edge profile with discretionary setups has been better than mechanical over same period in my testing so far, so it has value too + there is room to get better.
So, both can help each other. If we do it haphazardly obviously it wont work.
Pure discretion is hard to do. Adding systematic elements to it is much easier. In another forum, many traders use a mix of systematic and discretionary approaches, with more discretionary in trade selection and more mechanical in other things ( filter/Trade management/position sizing ). So maybe 10% discretion + 90% mechanical.