getting better

I have used the so-called Kelly criterion to good effect over the last 3-4 years with my sports betting.
However I wonder whether I can apply it to work out how much of my pension fund should be invested in equities?
If I assume an equity risk premium of a say 3%, and a log normal distribution of future equity returns, I seem to get 100% over whatever period I choose.
This conflicts with conventional wisdom that suggests that the longer the period the greater the proportion should be in equities?
Perhaps the utility function implicit in the Kelly criterion needs to be changed for this purpose?
Has anyone looked at this issue?