CFAI has an interesting document on the whole process from their perspective. Ctrl+f “70” didnt find about the 70% rule of thumb in this document so I think its safe to say its not a current practice.
http://www.cfainstitute.org/programs/cfaprogram/Documents/cfa_program_theory_meets_practice.pdf
I couldve sworn I read it in that doc or a similar one straight from CFAI a couple years ago though…
^^Realistically, someone mentioned it on this forum 5 years ago and its been repeated since
I do remember that I had read a similar but another PDF file that says No on who scored 70% failed. Now that wording is gone. What does this mean? Scoring 70% is not guaranteed to pass. Can anybody else affrm that you had read such PDF before?
I am 100% sure i read it in July 2013 on CFAI website itself.
So, that means, even 70% is not safe. OMG
Just saw this score of a guy on 2014 thread and who passed L2
Q#TopicMax Pts<=50%51%-70%>70%-Alternative Investments18*—Corporate Finance36–*-Derivatives36*—Economics18-*–Equity Investments72–*-Ethical & Professional Standards36–*-Financial Reporting & Analysis72-*–Fixed Income Investments18-*–Portfolio Management36-*–Quantitative Methods18*–
Just had a cursory glance over this score, I guess this can be very very close pass indeed. Perhaps the MPS
Wow, this one gives me hope…
First post on AF, woot!!! I distinctly remember reading on the CFAI website that nobody has ever scored 70 or above and failed. However, I also remember the website pointing out that there is no guarantee that this will be the case going forward.
If someone were to get 70+ on all topics and fail (obviously an improbable circumstance, but still possible), they will probably post on here to . Let’s hope we don’t find a post like that come August.
As for the ethics adjustment, who really knows? I’m glad they don’t have a FI or FRA adjustment though…
http://www.cfainstitute.org/programs/claritas/Pages/claritas_faqs.aspx
However, considering the historical range of passing scores for the Claritas Program candidates who can achieve 70% on the exam should feel confident about their performance.
Claritas is just like L2 
If you score 70+ in all topics, then you most certinaly are above 70%, and hollistically, you are probably way more than 70%.
It’s safe to say the MPS is in the 60-70 range, based on aggregates of pass and fail samples.
I agree, just pointing out that this is the only way for sure to really know if the MPS went above 70. It would be devastating to get such a score. Plus the % of condidates that passed would surely be well below the historical average if that was the case thsi time around.
Its kind of nice to know know the MPS though–there are countless comments on trying to derive the MPS that have kept me quite entertained since the test.
Haha…it has kept us entertained…though on June 6, the discussion was whether or not MPS can be below 65%, now after 2 weeks, the discussion is on whether or not MPS can be below 70%
deleted, can’t be bothered to correct my calcs… and it’s sunny outside.( sorry lovelords)
Some miscalculations you have done:-
Economics wud be out of 6
Max possible score in less than 50% range is exactly 50% since CFAI designates this as <=50.
The MPS changes year-to-year. It’s based on the difficulty of the exam compared to previous exams. The idea is to balance the number of candidates passing with the same competency maintained throughout the years. No one really knows how it works. All we do know is that the pass rate is getting lower as the pool of candidates get bigger, and the passers are simillarly competent independant of the time the was exam taken.
This only makes half sense. If the idea is to maintain the same level of competency of those who pass, then the same % of candidates should pass each year. That’s not the case though. I’m not sure why CFAI doesn’t determine the MPS to smooth out the fluctuations in pass rates—with hundreds of thousands of people taking the tests they can’t argue that the level of competency (on average) changes from year-to-year.
As an anology, if 100,000 people went to take their drivers’ test and 75,000 people passed, you would expect that 75% of the people who took their drivers’ exam next year would pass, regardless of the number who tried. And however many who passed would be just as competent (as measured by reaching the minimum requirements of the driving test) as those who passed last year’s test.
CFAI basically tweaks the “minimum requirements of the driving test” each year. One year you’re allowed five moves to parallel park your car, the next year you’re only allowed four. Any apparently it takes about 8 weeks to do so.
I have no clue if anything I just typed made any sense, but that’s what i think of the whole MPS system.
CFAI states that they use the modified angoff method. Basically the method works like so:
I believe first and foremost, exam questions that had received numerous complaints (as to whether they were confusing or poorly formulated) in the 5 day period immediately following the exam are Investigated and credited to all candidates if they deem the question unacceptable, unfair, too confusing, or has more than one correct answer.
Then, they put together a group of experts who look at each of the remaining questions on the exam as single items and the experts are asked for the percentage of CFA charterholders they would expect to answer each specific standalone question correctly. Not a CFA candidate, a minimally qualified PRACTICING charterholder. Each expert comes up with a probability or p-value and then the p-values from each expert are averaged for that particular question. Values are capped (because they know it’s impossible for every charterholder to score perfectly on any particular question) and floored (because anyone could guess and answer correctly). A possible range may look like .33 - .90 for instance but your guesses are as good as mine as to what they actually use.
In the modified angoff method, the experts discuss what they decided after submitting their values and have the option to adjust their p-value. I’m not certain, but they may only do this for questions where the standard deviation of the p-values is above a certain threshold (i.e. 10%). Once the experts have finalized their p-values, they find the mean p-value for that question. So a question may have an average p-value of .738, for instance, meaning the committee believes that 73.8% of charter holders would score correctly on that particular question.
Once all of the mean values for every question are determined, the means are added up and once again averaged. They then round either up or down to make it a whole number. This figure is the MPS for that exam.
I could be wrong on a detail here or there but from my understanding of what I read about modified angoff, that’s basically the jist of it.
So basically, if you want to try to estimate the MPS for any particular year, you’d have to guess what the committee of experts thinks about each question. This could also explain why ethics is such a crucial section despite its actual weight on the exam.
If anyone has a better understanding of modified angoff, please correct me if I’m wrong anywhere.
Source: https://www.questionmark.com/us/seminars/Documents/webinar_angoff_handout_may_2012.pdf
Source: http://www.cfainstitute.org/programs/cfaprogram/exams/Documents/exam_results_timeline_june.pdf
Some of you may notice this source is the same company the powers CFAI’s online practice test questions. The Questionmark Perception product is fully explained on their website. It’s interesting because it also provides CFAI with analytics for each question. However, the extent to which they use this product is speculative at best for us candidates. If I had to hazard a guess, I’d say they’re currently testing this software which could mean they modify the way they test us, analyze our results, even the frequency or ability to test candidates in the future. The conference centres with thousands of people and the growing number of candidates I’m sure poses a problem for the growth of the program and let’s face it, education is a business. And business is a-boomin’.
Source: https://www.questionmark.com/us/perception/Pages/default.aspx
Careful using this one, it has weird formatting. The “max” passing score someone calculated from this score last year didnt omit the dash at the beginning/end of each entry for the 63% number. This score could actually be as high as 77%. That does assume he went 24/24 in EQ and 12/12 in ethics.
of course, he could’ve scored as low as a 53%.
One other note in the estimation of MPS… We don’t know which questions get dropped. All of the previous speculating assumes there are no dropped questions. For instance, if a candidate scores 3/6 and one of their correct answers is dropped, it would become 2/5. Unless of course they credit candidates instead of dropping the question. Does anyone know for certain?
edit: answers are credited, not dropped. Never mind.
So do you guys think the MPS is always a flat number?
Or could it be something like 66.67?
It changes every year based on the difficulty perceived by their scoring committee. Check out my above post.