Actually, one cannot give a universal answer to this question. I think, one of the main parameters to optimize are the primer concentrations and the ratios of outer and inner primers. You will have to do optimize each assay, which is only worth when planning to do a lots of samples with that assay. Otherwise, I would recommend switching to another method, if possible.
First of all, you have to know what are the melting temperature of each primer. After this, you are able to start your experiment design. In the Tetra-primer ARMS technique, the primer balancing is one of the most important steps for having good results. Ye et al., indicates that a proportion of 1:10 for each outer primer/inner primer is the best balance. Other important step is the melting temperature used in the PCR cycles. You have to favor the melting temperature of the inner primers (It is suggested that the inner primer have a melting temperature lower than the outer primers). You can also, change the magnesium concentration of your PCR buffer and try some kind of touchdown PCR if the standardization of your PCR is difficult. I suggest that you read the Ye paper for mare information (http://nar.oxfordjournals.org/content/29/17/e88.full.pdf+html). But, the most important step in the etra primers ARMS-PCR is the primer design. You have to be sure about the specificity, location of the miss-match and the 3' pattern in your inner primer. You can also read a paper that I published where the Tetra primers ARMS-PCR was used to genotype a polimorphism in the bovine kappa-casein (http://www.funpecrp.com.br/gmr/year2013/vol12-4/pdf/gmr2523.pdf).