CT8 September 2018 – Predictions

Discussion in 'CT8' started by Dom B, Oct 2, 2018.

Thread Status:
Not open for further replies.
  1. Dom B

    Dom B Member

    As many of you know, I commented extensively on the April 2018 paper as the pass rate was so low at 31.6%. Most students who sat this paper acknowledged that it was abnormally tough. A minority of more disingenuous folk, who passed an ‘easy’ sitting of CT8 insisted that it was no more difficult than any other paper.

    So how did the September 2018 paper compare with April 2018 and what are the likely pass rates this time? To predict pass rates there are 3 primary factors to consider:

    1) The relative difficulty of the paper;
    2) How many students are sitting the paper;
    3) What is the quality of students likely to be in this sitting?


    1) The relative difficulty of the September 2018 paper
    My view of the question difficulty on the September 2018 is as follows:

    (i) Very straightforward questions (49 marks)
    1. Behavioural finance – this is a gift of an opening question and most students will pick up close to 10 marks here
    3. APT – a nice easy question, 5 marks
    5. Ornstein-Uhlenbeck proof – straight out of the core bookwork, guaranteed 8 marks
    6. Option pricing and proof of put-call parity – straightforward enough, most students will capture most of the 10 marks here
    8. A nice easy question on the binomial model – most students will capture all 7 marks here
    10. Black-Scholes – simple. Most students will score close to 9 marks

    Reasonably prepared students will score a minimum of 40/49 on these questions

    (ii) Intermediate questions (45 marks)
    2. Lognormal distribution & VAR – some sections of this question are straightforward, some are a bit tricky. But most students should score at least 5/11 here
    4. Lognormal distribution – most of this question is straightforward with some aspects being tricky. Most students will score at least 7/14 here
    7. Merton Model – parts (i) to (iii) are straightforward and (iv) & (v) are not that bad either – most students will score a minimum of 8/13 here
    11. CAPM – the way students score on this one will depend on whether or not they got tripped up on part (ii). Most will get part (i) scoring a min 2/7 but many should do better given that it is so similar to question 11 on the April 2018 paper (which most students will have practised)

    Reasonably prepared students will score a minimum of 22/45 on these questions

    (iii) More difficult questions (6 marks)
    9. This is a slightly more obscure piece of bookwork. If you knew it, you get all 6 marks. If you don’t, unfortunately you get 0 (much like question 7 on the April 2018 paper)

    Quite a few students will score 0/6 on this question but others will score 6/6

    (iv) Near impossible questions (0 marks)
    None in this category


    How this compared to April 2018

    · The April 2018 contained a 12-mark question (question 5) that was near impossible, taking all 12 marks off the table for almost every student. There was no such question in September 2018;

    · The April 2018 did not have any easy bookwork questions, such as the September 2018 10-mark question on behavioural finance or the 8-mark question for writing out the Ornstein-Uhlenbeck proof

    Verdict: This is CT8 paper is of average difficulty. It is not as easy, for example, as the April 2016 & April 2017 papers, but is certainly more straightforward than the September 2015 paper and far more straightforward than April 2018. Therefore, in a normal sitting, one would expect the pass rate to be somewhat close to the long-term average of 53%. However, this was an unusual sitting which will have an impact on the pass rate (see below).


    2) How many students are sitting this paper
    During the last 4 sittings, between 800-900 students have sat this paper. However, following the April 2018 horror show, there will be an additional 250 students re-sitting this subject than would usually be expected. So we are probably looking at around 1,100 students sitting this paper.


    3) What is the quality of students likely to be in this sitting?
    Generally, when you consider a sample size of 1,100 actuarial students sitting an exam, it is nonsense to assume that there will large arbitrary variations in the average quality of students from one sitting to the next. In general, Low pass rates = more difficult paper.

    However, in this sitting there is one important factor to consider: There are ~ 250 students re-sitting CT8 because they scored in the 50’s on the April 2018 paper. Almost every single one of these students would have passed the September 2018 had it been presented to them in April 2018 (or any other CT8 paper for that matter) and will therefore pass this paper.

    So if 1,100 students sit this exam:
    Roughly 90% of 250 re-sitters who scored in the 50’s on the April 2018 paper will pass
    Roughly 53% of the remaining 850 students will pass

    Verdict: Overall pass rate will be 61% (although could be lower if there are more than 1,100 students sitting this exam – i.e. more ‘first-time sitters’)


    Overall verdict

    Accounting for the difficulty of the paper, together with the fact that the quality of students sitting this will be higher than average, I expect the pass rate for this exam will be in the high 50’s to early 60’s.

    Interested to hear the thoughts of others.
     
  2. Hi Dom B, I spoke to quite a few students after the exam session, and many found questions 2 and 4 difficult. I am only expecting to get a few marks for question 2, but did feel better about question 4.

    Question 9 had me entirely stumped, as probably was the case for most others.

    But apart from these questions, I do think this exam was of decent standard.
     
  3. Dom B

    Dom B Member

    Yes I would expect students across the board to have a similar opinion. I think it won't matter that you only scored a few marks for question 2 if you scored well on the other questions. Best of luck.
     
    Last edited by a moderator: Oct 3, 2018
  4. Dom B

    Dom B Member

    So the record number of students taking CT8 in a single sitting was 918, but 1,734 sat the September 2018 paper.

    I predicted that only 1,100 prepared students would have sat this exam, and did not anticipate than an additional 634 would take a hail mary swing at this before it turned into two papers (which even for a reasonable CT8 paper like this is rarely going to pay off).

    So 720 of 1,100 passed giving an effective pass rate of 65.4%, which is slightly higher than I predicted.
    The remaining 634 have not been a feature of any previous CT8 sitting so can be excluded from the effective pass rate.

    Congratulations to those who passed, especially the re-sitters who had to suffer that April 2018 paper!
     
    almost_there likes this.
  5. student1990

    student1990 Member

    You could argue only 720 prepared students took the exam ;)

    I suspect a few more than 1100 were prepared using your definition as some people decided to give CT8 a good go instead of trying CT4 or 6 given the transition arrangements.
     
  6. Sam88

    Sam88 Member

    I decided to go for the hail mary swing and I passed on my first sitting! Congratulations on those who passed!
     
  7. Dom B

    Dom B Member

    Ha! Well I think that would be unfair to the ~ 400 students who prepared well but just b*llsed up a couple of calculations on the day

    Possibly more than 1,100 well prepared, but not more than 1,250 - which equates to a ~ 58% effective pass rate.
     
  8. Dom B

    Dom B Member

    Congratulations Sam. You certainly picked the right CT8 paper for a hail mary swing that's for sure!
     
    Sam88 likes this.
  9. Sam88

    Sam88 Member

    Can't say the same for CT4, that was a killer. CS2 in April :(
     
  10. Dom B

    Dom B Member

    Bummer. Nothing irregular about the number of people sitting that one but only a 39.4% pass rate so must have been very tough.

    Still - you nailed CT8 so make sure you sink a few beers this weekend!
     
    Sam88 likes this.
  11. Sam88

    Sam88 Member

    Cheers Dom! For sure on those beers!
     
    Dom B likes this.
  12. Interestec

    Interestec Keen member

    I don't see how you can just predict that effective pass rate was 58% from an actual pass rate of 41.5%. I was a first time sitter and passed and yet in my place of work there were second time sitters who still didn't pass. By your reckoning I shouldn't be counted as I was a 'hail mary swinger' and the odds were stacked in their favour. Realistically anyone who prepared well enough should be counted in the statistics - there could be people who've sat it five times and failed consistently, the only concrete evidence we have is that 41.5% of sitters passed. It still seems a low enough pass rate to me compared to historical average.
     
  13. Dom B

    Dom B Member

    The evidence we have is that the record number of students sitting this paper in the last 23 sittings was 918. Almost double that number (1,734) sat this paper which also just so happened to be the final sitting of this exam before this turned into a larger subject with two exams ;). As I stated in my original post "Verdict: Overall pass rate will be 61% (although could be lower if there are more than 1,100 students sitting this exam – i.e. more ‘first-time sitters’)"

    This paper was slightly more difficult than any of the 2016 & 2017 papers so you would expect a lower pass rate than those papers. But even the most disingenuous of students who worked their way through ASET would recognise that this was a much easier paper than September 2015 which had a higher pass rate of 49.5%. The difference in September 2015 was that there was nothing abnormal about the number of students attempting that paper and no incentive for students to 'try their luck'.

    You shouldn't feel that this diminishes the fact that you passed, passing an average CT8 exam is a good accomplishment. It is just evident that the overall pass rate has been pulled down by 100s of students who were just trying their luck on this occasion.
     
  14. Interestec

    Interestec Keen member

    I think you missed my point. I don't disagree that extra people brought it down I just don't think you have any certainty in saying what you think the effective pass rate was. The way percentages are done is that it doesn't matter on the nominal amount of people... But I 100% agree that there were a lot more that just went for it that weren't properly prepared.

    I don't, finishing my CTs in 2.5 years has been a good enough of an accomplishment.

    My main point however was that yes there was a record number of people and some evidently were trying their luck. But there is always people that sit for exams that aren't properly prepared. Yes there was a record number but due to the fact that percentages are on the whole amount you can't say with any certainty how much the pass rate was skewed.
     
  15. Dom B

    Dom B Member

    That's why I provided a range of 58 - 65%. I understand that it feels better to have passed an exam that had a low pass rate, but given the relative simplicity of the paper compared with September 2015 and the clear evidence that 100s of students were taking a swing, the effective pass rate will be well into the 50s.

    Well done.
     
  16. Interestec

    Interestec Keen member

    I could care less if the pass rate was 90%, wouldn't feel any better or worse. You are making predictions with literally no fact backing your argument other than oh there was a record number sitting the paper, not all of them were prepared. If an extra 1000 sat it and 500 were prepared and 500 weren't they wouldn't skew the data at all. Of course if it had been a harder paper the pass rate would have again been lower. However you have no evidence apart from guesstimation what the effective 'range' was.
     
  17. Dom B

    Dom B Member

    I can see you are getting quite frustrated about this but I'm really not sure why. If you read my responses I have applied a lot more information than you suggest in order to reach this conclusion:

    - Almost double the number of candidates sat this exam when compared with the previous record high out of 23 sittings;
    - This unprecedented surge in students sitting CT8 just so happened to occur in the final sitting before CT8 became two papers;
    - Based on past trends of ~ 900 students taking this exam in each diet together with an additional 200-250 re-sitters following the April 2018 horror show, one would expect ~ 1,100 students to have sat this paper. This indicates that a significant proportion of the remaining 634 students were just taking a swing at this paper. Nothing even close to this has occurred in any of the previous 23 sittings;
    - Any sample of students taking a swing at any paper will inevitably have a much lower success rate than the norm. When you consider that at least 37% of students were in this category, that will dramatically drag down the pass rate;
    - This paper was considerably easier than the September 2015 paper (which had a 49.5% pass rate), even the most disingenuous of students would not argue with that point.

    This is exactly why I caveated my original pass-rate prediction to state that it was dependent on the number of first time sitters.
     
  18. Interestec

    Interestec Keen member

    I've understood your points and I agree with them. I'm only getting frustrated as you seem to be missing my point. I agree that the pass rate was skewed by the extra amount of students taking 'a swing' at it. And I agree that the paper was easier than September 2015. I don't agree however that it so easy to say that the exam has between a 58-65% effective pass rate. I just don't think you can attach figures when you don't know. You don't know what percentage of the first timers were taking a swing to any certainty.
     
  19. Dom B

    Dom B Member

    Well it isn't easy to say that. However, if you break down what we do know:

    - Any large group of students have a similar average ability from one year to the next. That's why A-level, Degree results, etc. rarely fluctuate by more than a few %. The same of course applies to actuarial students;
    - So if there is a substantial variation in pass rates, this is almost always down to the paper being more difficult;
    - If analysis reveals that the paper is not more difficult, then there must be some other factor affecting the pass rate. In this case, there was one obvious factor that was not a feature of any of the previous 23 sittings (which we both agree is a factor)
    - In this case we also both agree that the September 2015 paper (I feel very sorry for students who landed that one!) was more difficult so September 2018 students with the same average level of preparation should have achieved a pass rate with a lower limit of at least a few % above 49.5%
    - I'm sure we also both agree that this was not the easiest CT8 paper in history either so that places an upper limit of 64%
    - This already puts us in a range of 52.5 - 64%
    - I am then factoring in the ~ 250 who scored in the 50's in the April 2018. Almost all would have already passed CT8 had they not been handed that paper in April and this would have elevated the 52.5% lower limit to ~ 61% (had 1,100 students sat the exam)
    - Allowing for a 3% margin or error either side gives a range of 58 - 64% (which is where the effective pass rate mostly likely resides)
    - Therefore, there has to be another reason why this expected range was pulled down and the most compelling evidence is the unprecedented surge in number of students incentivised by attempting to clear this exam in the final sitting.
     
  20. Interestec

    Interestec Keen member

    Is it not reasonable that any one (or more) of your assumptions could be wrong pulling the effective down to say 57%. Or due to the factoring of the extra students from April who would have passed could they not have pulled the effective rate up to higher than 64%? Maybe there is a less obvious factor.
    In your assumptions you assumed 3% difference from 49.5% and then again assumed a 3% margin error on both sides. Is it not plausible that the difference is a different figure and say the range was 51-62% based on the same logic. Or even 53-64%. You have two many assumption and unknowns to (in my opinion) attach numerical values too (if even in a range).

    Could one not also make the argument as follows: The biggest data set we have is the cumulative pass range for CT8 over the previous years. From April 2007 to April 2018 16820 sat for the exam and 8842 passed giving an average pass rate of 53%. Now even the most disingenuous of students wouldn't believe that the paper was harder than average or in fact easier than average. Perhaps the paper was more along the middle thus tending towards the historical average of 53%. Now we could allow for people that would have passed given they weren't subjected to April 2018's paper but then with that allowance and the allowance for the people that just gave it a go as first time sitters and weren't prepared would we skew it upwards or downwards.... One would be inclined to leave it more in line with the historical average which has had 16x the amount of sitters. Even allowing a 3% margin error on this figure would give you a completely different range to the one you quoted.
     
  21. Dom B

    Dom B Member

    Well that's barely any different to the 52.5 - 64% range I formulated based on logical assumptions.

    Yes. As per my original post.

    The flaw in this argument is that you are assuming equal weighting on both groups. But based on historical data and the obvious incentive to clear this paper before it becomes two papers:
    - There are ~ 250 re-sitters who were already pass standard students and they will drive the pass rate up
    - There are ~ 634 students just taking a swing at this paper and they will drive the pass rate down
    So the numbers driving down the pass rate exceed the numbers driving it up by a factor of 2.5.
     
Thread Status:
Not open for further replies.

Share This Page