1.  More than a decade of experience we've condensed into a two-day course on the theory of linear statistical modeling and its actuarial applications with a focus on modeling loss triangles. In 2005 we presented this course four times at three reinsurance companies.  The linear statistical model is a powerful and practical tool for the prediction of multivariate means and variances; in fact, one valid statistical model is worth more than any number of deterministic methods.  This course is much less expensive than statistical software, and lets (re)insurance companies be self-reliant and creative.  Course syllabus.

2.  For years we've championed utility-theoretic pricing, as opposed to risk-adjusted discounting and return on equity.  Let the curious open the following two files for a presentation at a CAS Limited-Attendance Seminar in September, 2005:

Utility-Theoretic Underwriting.pdf

Utility-Theoretic Underwriting.xls

It may be difficult to understand the files without the commentary; however, the ideas are developed in The Valuation of Stochastic Cash Flows, which you can download at www.casact.org/pubs/forum/03spforum.  Naturally, it'd be our pleasure to talk to you about this, the science of risk taking.

3.  Many actuaries attempt to make statistical models out of the Chain-Ladder method.  We’ve written an in-depth critique on the bias of this method, which appeared in the Fall 2007 issue of Variance.  You can download here a copy of Chain-Ladder Bias: Its Reason and Meaning.  From it you will learn the true reason why this method commonly over-predicts, and will gain a new understanding of credibility.