We determine correlation estimates by matching the observed index option price with its model price. Different basket price approximations lead to different implied correlation estimates. The industry standard for measuring implied correlation is a bad estimate when it is needed the most. We propose a new approach for determining accurately implied correlations.
