How do you assess the quality of an ICT Breaker Block versus an Inversion Fair Value Gap?

Both setups utilize failed structural levels for entries. Which of these two concepts has provided a higher win rate and cleaner execution in your historical journal?