April 2008

I was lucky enough not to have to suffer the stress of those who took the exam in Barbican... the paper I thought was pretty fair - easier than the 2 papers from last year. Still touch and go for me however...fingers crossed.

How was it for you guys?

 

Unfortunately, I sat the exam in Barbican. The whole shananigan was a utter shambles. I was exhausted before the exam even started.

 

Eek... looking forward to a FA unfortunately... You're right, exam was MUCH easier than previous years but I forgot some stuff that I should know.. for example, I used the wrong formula for the Kaplan Meier, and I forgot how to do the proof as well as the entire ques. num. 11. Utter disaster.
Well, CT2 next.

 

How did you guys lay out the state space for the audit firm question? What was actually asked in part c) - the prob that company use each firm in long time?

I ended up with 9 states (ij where j is the firm used in current year and i is the firm used in previous year). I figured that I can never work out a stationary distribution!

 

I thought the exam went well but I was quite prepared also.

For the state space question, I had six states. A_1, A_2, B_1, B_2, C_1 and C_2.

Where A,B and C is one of the firms and the subscript is the year of being in that firm in just that contract term.

This is Markov and has discrete states and time space... so hopefully it was what they wanted!

The transition probabilities also went quite easily, i.e. just from A_1 to A_2 and then A_2 goes back to A_1, B_1 or C_1.

Then to answer part c, since this is irreducible and aperiodic. I worked out the stationary distribution, which boiled down to only three equations. And then added up X_1 and X_2 to get the long run probability of being in state X.

I think I made a silly mistake on the exposed to risk calculation which might of screwed up the answer, but besides that the paper went well.

What went wrong in the Barbican?

 

If I'm not wrong, that works out to be the same as the stationary distribution if you work with 3 states (A, B, C) alone.

Well, it was a nice paper.. I hope the marking will be equally nice as well...

 

Yeah now I am actually thinking it could of been right just to use three states and have the time set as 2t, t=0,1,2....

It comes up with the same answer and is easier to work with.

 

Mate you've made me worry beyond belief, I didn't see anything in the question that indicated extra states being made! I thought the amount of computation required was worthy of the marks being offered too!

I thought it was a decent paper which was doable, but totally forgot how to do Cox Regression and Question 11. I only remembered in the last minute and eventually got the likelihoods down, was gutted to miss the proof as I had done something similar the night before! FA for me, however considering the Barbican delayed start, maybe the pass mark will be brought down to 45 for that! Here's hoping!!!!!!!!!!!!!

 

how did you do it?

for question 11, was that the last one?, the proof was show the forward differential equation from first principles?

What about the internet service provider question.. that was a little tricky also!

 

I could be wrong though mate, I wouldn't say that i'm the most knowledgable when it come to exams now! I did the MArkov question with 3 states {A,B,C}.

The ISP one I only did the beginning, I started doing the differential equation but forgot and seperating for the integrating factor, but can't remember why I didn't carry on. I think it was because when I integrated I didn't have anything to calculate the constant with.....dunno!

That was question 11, I started the proof, but for the life of me, couldn't think of how to move on with it. I ended up doing the likelihood and then ran out fo time. I had like an hour left to do questions 9 and 11 but suffered brain freeze!

 

What do people reckon the pass mark for this exam is anyway?

 

The Markov question could definitely be done in three states. I did it the more complicated way but I am pretty sure we end up with the same answer.

The telephone question had the initial condition that you start at off and that gave you the constant.

The proof on 11 was so annoying. I started it thinking this is going to be an easy five marks.... then I ended up with a missing term.

But luckily something clicked and I got it. The term was going to dead then staying there.

 

I was gutted mate, I think it was passable paper, but looks like I might be resitting for the 3rd time in September!

 

Hi

I also did the Markov question using 6 states the same as Fiend. I also had six states. A_1, A_2, B_1, B_2, C_1 and C_2.

Where A,B and C is one of the firms and the subscript is the year of being in that firm in just that contract term.

I had a different set of transition probabilites than fiend though because I read the question that the conract were rolling, i.e. the minimum 2 year contract continued and as such I had the transition probabilities, from A_1 to A_2 and then A_2 goes back to A_2, B_1 or C_1. But that probably why i could not solve my stationary distribution.

Oh well always september!!

 

Wow, at least I'm not the only one suffering... The pickle is that since it wasn't such a bad paper, then Sept might be harder, and the pass rate will be high for this paper...
Here's to high hopes.

One Day We'll All Pass.

 

Does anyone else find this post exam analysis painful?!