Sorry since that was a little confusing. Basically: The sensor will increase the output level when tilted anywhere from level (the point at which output is zero). My idea would involve two channels, one for left and right. The signals for both the left and right would be equal in volume when the guitar is in playing position. When the guitar is tilted, the volume for one channel would decrease, producing the fading effect.
My first thought was to have the sensor wired so it would simply bypass the signal for a channel to ground as it was tilted. That is, the sensor would be mounted so it would be level at playing position. The output of the sensor would be wired to ground. As the sensor is tilted, its output would increase, thus filtering more of the signal to ground, reducing it in volume.
However, the sensor increases in output as it is tilted either left or right. That means a channel would decrease in volume when tilted anywhere away from playing position. This would be the case for both channels, so tilting the guitar any way would decrease the volume of both channels, which would not be a fading effect--rather a volume control.
In order to remedy this problem, I would use a mercury switch: The sensor in the left channel would be connected via a mercury switch. The only way for the sensor to filter the signal to ground (to reduce the volume) would be when the mercury switch is closed. The mercury switch would be mounted so it would be closed only when the guitar is tilted to the left. That would stop the left channel circuit from reducing its volume when it is tilted to the right, as the mercury switch would then be open. The right channel would be wired in the same way except the mercury switch would be closed when tilted to the right. Effectively, wiring the circuit in this manner would allow each channel to be reduced in volume only when tilted one way, solving the original problem and producing a fading effect.