You need to normalise the encoder position so that it stays in the range between zero and however many pulses make up a complete revolution. Assuming the encoder is arranged so that the normal direction of rotation increments the position, you can do that at the point of incrementing it with the modulus operator like this:
encoderPos = (encoderPos+1) % PULSES_PER_REVOLUTION;
Then you can work out which sector it is in by division:
sector = encoderPos / PULSES_PER_SECTOR;
Then you can use sector as the index into your array of sensor pairs, or whatever else you want to do with it.