CT4(103) - Sept 2006

Could anyone draw that transition diagram for Q5 - the one where it depends on claim record in previous two years? I've never seen anything like it before!!

It was the only question I really couldn't do (except one part of last q).

Seems a bit unfair though 'cos the rest of q5 looked pretty simple, but if you can't get your transition graph down you were pretty much locked out of the entire question!

I think I got between 47-51% so will probably get an FA, unless I'm lucky!

What did you think?

 

I did the whole of CT4 and agree that Q5 from 103 half was tricky
In the end I had four states 0,1,1-,2
did it in a rush at the end (came back to it) and hope I guessed it right!!!!

 

Just the way it was written - I had real trouble understanding what the state spaces were and how they related to one another. Normally I'm really good at transforming a non-Markov chain into a Markov one - but not this time.

I think your answer looks right - I guess it was probably really simple and the examiner will be surprised if a lot of people get it wrong.

Just hope I did well enough on the rest of the q's.

Does a Poisson process have a continuous state space? I thought it did as claim numbers can theoretically go to infinity.

 

yes claim numbers can go up to infinity....but they can only take discrete forms..i.e. there can only be 1, 2, 3 or say a fixed number....claims numbers cannot be continous like 1.2 claims etc.

i may be wrong.

 

Thank you for taking time to answer.

I realise they can't be 1.2 etc - but S = {0,1,2,......Infinity}
Is that a continuous S or a discrete S ?

I do realise that this is a pretty remedial question which is why I was surprised I even needed to think about it - but the exams can do strange things to our brains!!

 

It is discrete. The state space is the set of all integers Z - which is discrete.

F*** !!!

 

Ct4 2006

I struggled with that one too! Thats always the problem with those questions - I had the same problem with the NCD model for CT6 - because I couldnt do the original transition diagram, I could not derive the transition probabilities, hence couldnt calculate the steady state etc!

On speaking to people afterwards, it seems that you start with 3 states and then extent it to 4 states somehow to make it Markov! In the way that you would complete the NCD model with the 3+ and 3- states etc. Im not convinced though either!!

Did you manage to do the question about how the recent mortality tables were produced?? I swear thats not in the notes?? Also the 0.25P80 thing using the ELT15 tables - I thought that was a CT5 problem??

I think I got at least half marks, so it will depend on the pass rate as to if I pass I suppose.

 

CT6 sept 2006

I struggled with that one too! Thats always the problem with those questions - I had the same problem with the NCD model for CT6 - because I couldnt do the original transition diagram, I could not derive the transition probabilities, hence couldnt calculate the steady state etc!

On speaking to people afterwards, it seems that you start with 3 states and then extent it to 4 states somehow to make it Markov! In the way that you would complete the NCD model with the 3+ and 3- states etc. Im not convinced though either!!

Did you manage to do the question about how the recent mortality tables were produced?? I swear thats not in the notes?? Also the 0.25P80 thing using the ELT15 tables - I thought that was a CT5 problem??

I think I got at least half marks, so it will depend on the pass rate as to if I pass I suppose!

 

The state space is discrete.

Hope this helps. No hope for me, I'm sure I failed

 

The only good thing I've heared about ct4 (from a selfish point of view) is that everyone who has commented on it reckons they've probably failed.

Hopefully this will result in a small lowering of the pass mark to say 47% so a few of us who think we've failed actually pass!

What is the normal pass mark anyway? I'd guess about 50% ?

 

oops didn't realise you'd answered your own question.

Am I the only one who was pressed for time?

 

Time wasn't an issue for me - but then I was locked out of the entirety of question 5 'cos I couldn't do the first part - so you could say I had a bit of time on my hands!

Thanks for answering my question - I edited it with the answer at about the same time you posted yours. I actually swapped my answer for that question about 4 times in the exam, I should gone with instinct over brain - then i would of been right!

When will I learn that in times of in-decision go with your instinct and not your reasoning?

 

Unfortunately I seem to be prone to "times of in-decision" during exams. It is as if a gremlin hiding in my head hides all my (limited) knowledge!

I also had a real issue with time ..... it was going okay until I came across the state space question. I spent a lot of time thinking about it but didn't crack it. Then I was under real time pressure for the rest of the exam.

 

Ct4 (103)

I had a real issue with time. Main reason was because of there being too many "thinking" questions. At first glance I couldn't do the first part of the first question so i left it to return later but never had time. When I eventually got to the last 2 questions I only had 30 mins left. I think i cracked the markov chain question (used 3 states to represent whetehr someone has: claimed in last 2 years, claimed in last year or not claimed in either of the last 2 years. However spent too long on the question at the expense of the 8 mark part of the last question. Overall, the paper was far tougher than the April paper. I was sure i'd passed that and it went well but obviously not the case. This paper went considerably less well but hoping a lower pass mark may well prevail.

 

I'm fairly confident I have 60%+ on this paper. I'll be well gutted if I fail it! Though last night I did think back to one or two to the smaller mark questions and realise I got them wrong, such as B1 but surely with so much to do in just 3 hours then people are gonna make silly mistakes. However I think the big mark questions are in the bag.

 

ok for q5, this is how I did it

start off with a state space, call it 0 (zero).
if you claim, you go to state space 1
if you claim again you go to state space 2
if you don't claim you go from 1 to 1-
you can go back from 1- to 1, if you claim
you can't go from 1- to 2, as you must go through 1, but you can return from 1- to 0 if you don't claim

for you to be at state space 2 then you must have claimed for last 2 years. you get to state space 2 from either 0->1->2, or 1- -> 1 ->2. if you're at state space 2 and you claim then you stay at state space 2 with the probability 0.25. If you don't claim you can go to 1-.

if you're at state space 1 or 1- then you must have claimed just once in last 2 years. to go from state space 1 to 2 you need to claim and it's with probability 0.15. Also to go from 1- to 1 is probability 0.15

if you're at state space 0 then you must have not claimed for last 2 years, so probability of claiming and moving to state space 1 is 0.1

you can fill the rest in

 

Yeah, I don't remember reading that either.

For the .25P80.5 question I had a guess and said since the mortality rates were assumed constant over the year, linear interpolation could be used first to find 1P80.5 and then .25P80.5 (since 0P80.5 = 1). It was just a stab in the dark though so it's probably wrong.

I did the same thing as Scarlets, who is confident of a pass, on A5 too so that's good to know!

 

I was lying in my bath this morning and all of a sudden I knew how to answer that state space question, it is so simple!

Isn't that always the way - if only exams allowed you to return within 24 hours.

 

im starting to think my answer was wrong now

i simply put in 3 states (0,1,2) as i thought this solved all cases. If you are in 0, you had to come from 1 or be in 0 already, so therefore you couldnt have claimed in last 2yrs. If in 1 you had to come from 0 (i.e 1 claim in last 2yrs) or from 2 (1 claim and 1 no claim). If you are in 2 you had to come from 1 or already be in 2.
Does this work?

Also i talked to one of the tutors who was an ex examiner and he told me that the pass mark for CT exams is 50-60 with an average of 55. The mark goes up and down according to how much people answered and not how well they done. The marks are also not weighted.

 

Isn't this thread in the wrong place? ("Exams" rather than "Subject" discussion.)
(Note from Admin: moved now - thanks for pointing it out!)

I have also heard that 28 marks out of 50 should secure you a pass.

It's interesting that there's a comment on the CT4 section that says that "the last 2 questions on the first part were difficult". Makes you feel a bit selected against only doing the CT4(103) part. Again.

I don't think there are enough practice questions available for these unusual state space ones the Institute obviously have taken a liking to. Weren't there only about 3 marks available (about 4.8 minutes) to come up with transition diagram and matrix? Way out, I'd say. I came up with 5 states, but I think it could collapse to 4, as per Scarlets. 3 doesn't seem enough. Hope I'll still earn marks for the later bits though.

I wouldn't like to predict the outcome on this one either. I think I should get it, as completed most of the rest of the paper, but I've been wrong before and neither outcome would surprise me.

 

I'm kicking myself for getting B1 wrong, it was uniform distribution of deaths wasn't it, I got confused thinking it was mu as constant so I've lost 4 marks there, dammit