Just looked at some notes on it:
The first point is that he had disconnected the autothrust because it would have got in the way of what he was intending to do. This of course disabled alphafloor, so that doesn't come into it. He was then flying height using elevator and thrust control was manual. In other words he had complete authority over energy management.
Then he was attempting to demonstrate maximum lift, but the system, sensing a deceleration (he was by now at flight idle) limited the aoa to about 16deg.
Some have argued that if this limitation had not been there he could have pulled a few more degrees (somehow without exceeding alphamax) and cleared the trees.
People have also queried why, with additional back stick demanded, the system actually applied down elevator. The answer to that latter is linked to the explanation of the accident - he had just applied full power and the nose-up pitch from the extra thrust, which is equivalent to a lot of up elevator at 110kt, was producing a nose up pitch acceleration which the system calculated would take the aircraft well above alphamax if not corrected.
But the essential point here is that he completely underestimated the spool up time of those fan engines starting from flight idle, and applied power too late to gain the required height.
Modeling showed that if the system HAD allowed him to pull up to alphamax he would have gained about 5ft at the treeline, which might just about have been OK, but because the thrust was not there to support the extra drag he would simply have sunk down again and gone into the trees a few yards further on.
The primary reason for the crash was that he applied recovery thrust about two seconds too late.