Could you clarify
- Do you want a single number for the moving window
- Do you want an output raster for the moving window that would have an average per sequence
This demo will exemplify the difference...(I have used numpy arrays to demonstrate)
Of course this can be greatly simplified is you use RasterToNumPyArray—Help | ArcGIS for Desktop to get the rasters into array format then use NumPyArrayToRaster—Help | ArcGIS for Desktop to get any needed back to raster format. Mean operation can be put through a 'moving' sequence so that you don't have to manually perform any of the steps.
So the result you need is important
import numpy as np
r1 = np.arange(9).reshape(3,3)
r2 = np.arange(1,10).reshape(3,3)
r3 = np.arange(2,11).reshape(3,3)
print("\ninputs\n{}\n{}\n{}".format(r1,r2,r3))
print("\nmoving average 1 {}".format(np.mean((r1,r2,r3))))
print("\nmoving,focal average\n{}".format(np.mean((r1,r2,r3),axis=0)))
Which yields
inputs
[[0 1 2]
[3 4 5]
[6 7 8]]
[[1 2 3]
[4 5 6]
[7 8 9]]
[[ 2 3 4]
[ 5 6 7]
[ 8 9 10]]
moving average 1 5.0
moving,focal average
[[ 1. 2. 3.]
[ 4. 5. 6.]
[ 7. 8. 9.]]