There are many sources giving different data on the initial position and the dimensions of the rods, but most suggest that the absorber was located just outside the core when AZ-5 was pressed, and the displacer graphite was located roughly in the middle of the core. Then the control rod begun slowly lowering, as can be seen here:
What happens next is puzzling, and I couldn't find a good explanation of it. According to the illustration, there are two reactivity decreasing zones ⊖, and only one reactivity increasing zone ⊕, the latter stemming from the graphite absorbing fewer neutrons than water while still lowering their speed. But this zone looks roughly equivalent to the second ⊖ zone, which is the inverse of it.
So in order for this to lead to a positive reactivity surge, aka positive scram effect, aka end-rods effect, there should be a lot more of reactivity going on in the bottom of the core, but it seems that the opposite was the case; in fact, more neutron flux was recorded on the top (see the ✱ line):
So why did the surge occur? Could it be that a lot of xenon-135 accumulated in the middle section of the reactor, and the bottom was relatively free of it, allowing for an unchecked reactivity increase in the presence of the graphite?
Illustration source: INSAG-7, pp. 123, 122.