Estimated Passing Cutoff - any comments?

[shauncass]

Assumed number of registrants who took the exam
20,000

top 5% would mean 1,000

Questions 140

Estimated passing cutoffs Estimate of average of top 1000 passed registrants
(no of questions out of 140) 90% 85% 80% 75%
if cutoff X = 75% 75% 95 89 84 79
if cutoff X = 70% 70% 88 83 78 74
if cutoff X = 65% 65% 82 77 73 68
if cutoff X = 60% 60% 76 71 67 63

Most probable scenarios ???

 

[mdkula]

How is the pass mark set?

 

[shauncass]

Pls see the attachment...

GARP SITE:

Will the new FRM Exam Levels I and II examination results be graded or treated differently from the FRM Exam Full?
The FRM Exam Levels I and II examinations will not be graded any differently than the FRM Exam Full. Candidates will not be provided individual Exam scores, the results will be communicated only as pass/fail. The passing scores will still be determined based on a ratio of the absolute score to the average of the top 5% of candidate test results. No penalty is assessed for a wrong answer. We will continue to provide you with an analysis of your scores as compared to your peers that took the FRM Exam.

And from David:

1. pass rate = X * Top 5% of (nbTakenTheExam); e.g., 0.7 * 5th quantile of (nbTakenTheExam)
which, of course, translates into some ratio of (nbPassed)/(nbTakenTheExam), which is likely to be less than 70% or 75%.

and again from David..

Thanks for updates, please keep them coming (please note: the FRM makes almost everybody “feel” like they failed; many have perceived they failed coming out of the exam, but passed anyway). @prem: GARP says they take the top 5th quantile (i.e., whatever score cutoffs the top 5th “percentile”), then they take a ratio (e.g., 0.75) of that top score. I personally suspect the ratio (0.X of top 5%) is calibrated after the results are tabulated to approximate a ~50% pass result; or, best case, they use .75 of top 5% (e.g., if 0.75* top 5% only passes a small % due to high difficulty, I doubt they would keep it).


But, either way, it’s really based on an (ex post) “grading on a curve” (no segmentation by section/topic, all questions weigh equally)...therefore: I do not think you can divine anything useful from the method. Unfortunately, it’s a waiting game - David

 

[shauncass]

Assumed number of registrants who took the exam
24,000
top 5% would mean 1,200
Questions 140

Estimated passing cutoffs Estimate of average of top 1200 passed registrants
(no of questions out of 140) 90% 85%
if cutoff X = 75% 95 89
if cutoff X = 70% 88 83