Just throwing out a random thought -
If you assume normalized IQ score is strongly correlated with ratio IQ score and then use the normalized IQ score to come up with an "equivalent age" then maybe you could make an (easily disputed) argument that this age would correspond to the maximum appropriate age mate group (assuming that the child's raw intellectual ability exceeds the child's achievement, social maturity, study skills etc.).
For example, if your child has an IQ of 140 and is 10 years old:
(equivalent age/10) * 100 = 140
equivalent age = 14
So an upper bound for acceleration would be four years.
I'm not actually suggesting this be used in practice. I think the other responses are correct. Obviously few 10 year-old children with 140 I.Q.s should be grade accelerated four years and some 10 year old children with lower test scores might be ready for extremely advanced material in some areas. Also, it seems unlikely that you would want to grade accelerate a child with an IQ of 110. But it might be just one way to start to think about an upper bound.
