I read the full report from the link Panther posted. Besides the typical upper echelon command CYA catch-all's in the report, it seems that like past accidents, there was a culture of arrogance/overconfidence and non-compliance to regulatory guidelines regarding both regular C-17 flight ops and demo envelopes. More than one C-17 pilot has flown the type in this manner and this was the only time the overconfidence came back to bite them. The pilot was well qualified having 3,251 total C-17A flight hours including 974 Instructor Hours and 124 Evaluator Hours.
Generally, the Mishap Pilot had taught student Demo Pilots to ignore stall warnings as they were "erroneous" in various parts of the demo envelope. Also, many standard procedure flows/ "Call Outs" were routinely pushed aside in demo flights. It should be noted that no two C-17 demo pilots adhered to specific procedures/flows.
The normal bank limitations are 45 degrees and no more than 40 degree nose up pitch initial max climb profile to 1500 ft AGL. These parameters were violated with the Mishap Pilot bunting over the nose at lower than 1500 ft AGL and utilizing 60 degree banks.
Also contributing to the accident was the retraction of the flaps and slats "on speed". The clean wing configuration led to a high loss of lift during the final turn.
Last but not least, the C-17A has both a Stall Protection System and a Deep Stall Protection AoA Limiter System(ALS). The ALS when active will limit the command nose up elevator position.
In this accident, when the ALS became active, the elevator surface movements decreased lessening the effectiveness of the Mishap Pilot's aft stick inputs.
The part about the ALS system bothers me. I will be willing to bet there are major issues with that system both Boeing and the Air Force do not want to admit publicly. Albeit the Pilot screwed up, the other factors involved go to show that safety automation are not fool proof and are not perfect by any means. The old saying I remember well: Automation is good enough to kill you!