In theory (only), it sounds reasonable because what you are effectively describing is using MAs to firstly define the direction and then secondly to act as a momentum indicator measuring them as they first diverge and the converge, I.e. measuring their difference as a histogram.
The problem here is that it would only work consistently if markets always moved in nice smooth sine-wave style waves. But that is not the case. Sometimes a trend will develop slowly and over a prolonged period and then end with a sudden drop over one or two bars. Conversely,sometimes the market will surge on a news release and then stand still and reverse back without any follow-through. Trouble is your MA differential will not react in the same way as it always requires the average of the same number of historical bars.
In other words, it will work whilst the market cycles in synch with your choice of MAs but will fail immediately when the market moves sharply or erratically.