For a mech system, only thing to do is to take all trades. Any change should go through testing phase first.
If risk is moderate and max dd is comfortable, then maybe one solution is to do full automation so that you wont have to face day to day pnl.
If you are trading with aggressive size in order to scale up, then probably no option but to follow MM plan and scale down just as you scaled up earlier. This is the trade off.
For my current system in stocks where i take a lot of trades and have some discretion in trade selection, my plan is to reduce activity (number of trades) in extended dd phases when market is trading adversely. I do this by becoming more selective in my entry criteria. This is not a binary on off thing though as within DD period we could have days with very good context and movement and i might relax conditions a bit.
I am now testing a plan in Nifty which works best with smaller stops and large RR. So win rate is very low and streaks of small losses are the norm. For this, I will have to take all trades, there is no option as any single trade can easily cover full/most of the DD.
The best edges for the mind probably have say 50-60+% win rate with say 1:1R - without too much clustering of wins and dependence on large wins, but i have yet to find such an edge in my testing so far in my market-tf, maybe due to high taxes. 1:1 is never optimal for my systems ..
Have you measured this ? Risk reward ratio will obviously change win rate, so this may just be a psych thing if the effect is not real. Otherwise why not take all trades in this way ?
A short term drawdown is likely to be random and so any action by us is in response to randomness and will yield random outcome. A more sustained DD env may benefit from change in tactics perhaps, not sure. There is an 'equity curve'
trader who says this can work but i have not tested mechanically.