LIVE Chat and Q&A with Scott this Friday 12 Aug AEST, page-210

I believe you're referring to Scott's discussion on calculations for COS. He said that ERCE used 7 different parameters to figure out the COS, whereas internally IVZ only used 4. Each of the "parameters" was essentially a percentage converted to a value between 0.0 and 1.0 (usually less then 1.0). To calculate the COS they multiply all the parameters together, and because each parameter is a number below 1, the more parameters you multiply together the smaller the final value becomes, hence a lower COS. I can't remember the exact details but be believed some of the parameters ERCE were using effectively duplicated one another and hence lowered the final COS value artificially.