When I test any concept I compare the results for a set of standard exits, being a Trailing Stop (as a benchmark) and exits at the close of X periods, where X = 1, 2, 3, 4, 5, 8 & 13. I often find that exits 1, 2 & 3 periods offer the best balance between return, strike rate and drawdown, but I think that tends to reflect my preference whereas a trader with a different risk profile may choose a longer exit which tends to offer a bettter return but the strike rate is typically lower and therefore the drawdowns higher (due generally to longer exits delivering larger wins, but less of them because there are more frequent losses).
Trailing stops tend to produce returns that are comparable with returns for periods 2-3, i.e. good returns, but the strike rate tends to be a little lower and the max drawdowns therefore slightly higher. I haven't tried a trailing stop as you have suggested (min or max of last 3 candles), but I have tried trailing stops on the low or high of the 2nd last candle rather than the last candle. I have found that the return is slightly better than a standard 1 period trailing stop but the max drawdown is slightly higher. My guess is your suggestion would probably provide a slightly better return again but also a slightly higher max drawdown. I will see if I can modify my trailing stop formula for excel & test the 3 bar trailing stop idea & post the results to see if it matches my prediction (actually I will try 1, 2 & 3 bar trailing stops).
The other idea I have been experimenting with is a staggered exit, so rather than exit after 2 periods which is the length I usually choose, I have been trying partial exits of 20% of position after each of 1, 2, 3, 4 & 5 periods. It does seem to produce a marginally better return than a simple period 2 exit, but also a slightly lower drawdown - not many variations have a beneficial effect on both parameters.
