Welcome to TIA Exam 7 Review Course
Welcome to the TIA Exam 7 review course.
I’ve just published Books 2 & 3.
For each paper, we have the following material to help you study:
- Manual - explanatory document going over the concepts of the paper
- Lessons - video lessons, uses the manual as reference
- Exercises - Exercises for each paper to test understanding of the concepts
- Past Problems - Solutions to past exam problems
- Worked out Problems - 2-3 problems worked out for each paper on video
For the Review part of the course, we also have the following for each paper:
- Review Notes - typically 4 page document reviewing the important concepts in each paper. These are not explanatory
- Flashcards - flashcards going over formulas and lists in each paper
- Practice Exams - 3 practice exams
For ease of printing, we also prepare “Books”. These each cover roughly a quarter of the syllabus.
Book 1 - Weeks 1 & 2
Book 2 - Weeks 3 & 4
Book 3 - Weeks 5 & 6
Book 4 - Weeks 7,8 & 9
Finally, the forum is the place to post your questions. I’ll be reviewing it regularly for posts, and responding typically within a day or two. Last year there was a delay because I was working on preparing material for the new papers. This year only Shapland is new, and the material has already been posted. I’ll also be posting any travel I have planned in advance - when I’m on the road there will be a delay of a few days in answering questions.
Book 4 - I’m adding exercises, and worked out problems on video. These will be posted by January 25th.
-Alejandro
Thanks, this is very helpful. Are all of Books 2 and 3 new compared to last sitting? Or if I have printed material from last sitting, is only Shapland new?
Thanks!
I wanted to get your thoughts on what I perceive to be a misconception in the Mack (1994) paper. It surrounds the test provided for the variance assumption. Three possible variance assumptions are presented in the paper: (1) Uniform, (2) Proportional to losses-to-date, (3) Proportional to losses-to-date squared. While the choice of variance assumption does have a slight impact on expected losses (due to different weights being optimal in estimating the LDF’s), the primary impact to the weighted residuals will be the changing exponent on losses-to-date in the denominator.
Seems to me we should be looking for a trend in the spread of the residuals, not the residuals themselves. If the residuals go from positive to negative, that should suggest a problem with the assumption that future losses are proportional to losses-to-date. A problem with the variance assumption should present as a cloud of residuals that is attracted to the x-axis as you move right (suggesting you need a smaller exponent on losses-to-date), or a cloud that explodes away from the x-axis (suggesting you need a larger exponent on losses-to-date).
You could actually see this with the charts you had on the old manual that was posted here a month ago. The three Ci,3 Residual graphs on pages 11-12 all had a slight downward trend (suggesting a potential problem with the future loss assumption), but the second graph (Var prop to 1) had the slowest attraction toward the x-axis, while the third graph (Var prop to c^2) had the fastest. To me, this suggests the uniform assumption was the best of the three in that case.
Jay,
I’ve responded here.
http://www.theinfiniteactuary.com/mb/viewtopic.php?f=29775&t=35821