Looking at your example, I can see it would produce a number of behaviors that might be confusing. I’m not sure which ones are causing problems for you, so I’ll explain each of them in turn.
First, your first transition region has a transition timeline. I can’t tell whether your transition timeline has any content, but if it has none, the expected behavior is for your event to fall silent while the playback position passes through the transition timeline, i.e.: when the playback position transitions from the left instrument to the right.
Second, there is a known bug in inverted offset mode that causes the playback position to transition from the last value of the transition region to the last value of the destination region, instead of the first value as you might expect. Transitions from other positions on the transition region are not affected by this issue, so it only comes up when using transition region quantization. This issue is on our bug tracker, and will be fixed in an upcoming release of FMOD Studio.
Third, when a transition causes the playback position to jump to the very last position on a loop region, the loop region isn’t triggered. This behavior exists because many events make use of multiple loop regions arranged back-to-back, such that the endpoint of one loop region is in the same position as the start of the next; if each such loop region’s endpoint was respected when a transition occurred, then the last loop region’s endpoint being reached would cause the cursor to jump back instantly to the first loop region’s start. In the case of your event, this means that when the playback position transitions from the end of your transition region it arrives at the end of the other section’s loop region and immediately exists the instrument.