1. avatar says


    Thank you John for the detail lectures they are a great resource however could you kindly elaborate The Value of perfect information as it is not available in the course notes also there is no lecture on it. Would appreciate if you could kindly elaborate.

    Many thanks

    • Profile photo of John Moffat says

      It is part (c) of example 1 on Page 46 of the Lecture Notes.

      At the moment, the lecture does not deal with part (c) – I must re-record it – but the answer at the back of the Lecture Notes should make it clear.

  2. Profile photo of manonaseriousmission says

    Thanks as always John for your brilliance at conveying useful and intelligent information across during every lecture. This one yet again was superb! My comments now will perhaps be the longest you’ve ever had to read on these forums, so my bad.

    I have always viewed Decision Trees, amongst a few other topics, as Marmite – you either love or hate them. For me, it’s been a hate journey with DT until recently, when I realised that decision tree questions are best tackled by first understanding what in the world any management accountant are actually attempting to do with the concept in real life.

    The more I try to understand the concept rather than just chucking in numbers and probability figures as I read the question, the more grasp of the topic I’m getting. This particular video of yours has just reinforced my take on the topic, as you sort of showed at the beginning part of your working, that it is just best to read and somewhat understand the question first and then draw out the decision tree based on the paragraphs in the question, without being unduly caught up in the complicated bits such as the figures/probabilities just yet. To bug oneself down about the numbers too early on can complicate things and make one’s working a tad messy and confusing.

    Just as you have shown in this example, once a meaningful tree (comprising different courses of action as spelt out in the question) has been drawn, then the figures can be slotted in, making sure that one considers carefully what the proceeds are and what the costs are (I particularly like your idea of using a bracket right off in distinguishing the costs).

    Ultimately, I have found that the process of establishing a Decision Tree from Left-Right is as important as the backward working from Right-Left, because where the former helps to establish what alternative courses of action or outcomes exist, the latter helps with the actual decision through an assessment of what appears to generate a better result (most likely by way of better revenue).

    I am so grateful for the way you have taken the time to demolish (in a good way) this question for better understanding of the concept. I will continue to attempt a few more questions on this topic so I can be more confident on the day of exam, should it come out.

    Can’t thank OT enough

  3. Profile photo of Chris says

    Hi John,

    thanks for such a detailed lecture on this! One question i need to ask though is why don’t we take in the initial cost of the research at say points B etc?

    Many thanks in advance.

      • Profile photo of Chris says

        Hi John,

        thank you. Yes i appreciate in this case that the outcome is still the same however, had the results been closer together the $200,000 would have a bearing on which choice we would make…. wouldn’t it? I may be over complicating this but it’s just so i understand in my mind rather than just doing the question and assuming.

        Thanks again.

      • Profile photo of John Moffat says

        No – the actual final amounts would be exactly the same.

        Don’t forget that at points A and B in my answer, they are the outcomes if we do not do the research.

        Whether or not it is worth doing the research is one of the decisions that we have to make.

    • Profile photo of John Moffat says

      You can start like that if you want.

      But it would make no difference at all because from the ‘no market research’ branch you would immediately have another decision – ‘expensive’ or ‘cheap’ refurbishment.

      (The tree is only to help keep track of the various decisions, so for this bit of it you can draw it either way)

  4. avatar says

    Hi John. In the third paragraph of this example, (in EITHER case the probability of the refurbishment achieving a good result has been estimated to be 2/3). Why are we not using 2/3 on the ‘buy market research’ path?

    • Profile photo of John Moffat says

      Buying market research is our choice – so buying itself is not uncertain and therefore no probability is relevant.

      As to what happens if we do buy the research, the question says (in the next to the last paragraph) that the probability then of the result being good is 91%.

      (If the research was perfect, then if it said ‘good result’ then getting a good result would be certain. Here, the research is not perfect and so it is not certain to be ‘good result’ but it has a very high probability. Similarly, if the research says ‘bad result’ then the probability of the research being wrong and getting a ‘good result’ is 13%)

      I hope that makes sense.

  5. avatar says

    Dear John. I’ve sat only F4 so far and don’t remember whether we could use black pen only. Are we allowed to use different colours in F5 to make our graphs and decision trees more understandable?

  6. avatar says

    Pheeww! :P A bit complex I must say! Sir, I have one question. For the final answer you said: Buy market research and if the result is good do the expensive refurb. If result is poor then shut.

    I agree with the first part i.e buy market research and if result is good then do expensive refurb which will give us 7.43M. But if we shut we will only get 5M.

    How about we don’t do market research and get 7.17M, ain’t it better than getting 5M if we shut?

  7. avatar says

    Hi sir..just wanna ask….um can a decision point (the square) arises after an outcome( the circle) as the usual way is that in a decision tree it only has ONE decision point which is the square at the far left( the beginning) and a lot of outcome….

    • Profile photo of John Moffat says

      Yes it can – there can be several decision points. In fact it is more likely that there will be several and not just one decision point.

      (Maybe the first decision is which machine to buy, but then after a year or two we need to make a second decision as to whether or not to sell a machine and the decision can be affected by how well or badly the machine has been doing in the first one or two years.)

      The example in the course notes is an example of this.

    • Profile photo of John Moffat says

      Decision trees are a bit less likely in December because they were asked in the June 2013 exam.

      However, they could be asked. I cannot really add anything to what I say in the lecture, but if you say where you have the problem then I will try my best to help!

      • Profile photo of adejumolu says

        Thank you very much Mr. John Moffat. You are one of the best teachers I’ve listened to. I just don’t understand when to start calculating the expected values. I pray it doesn’t show up this December.

  8. avatar says

    The problems with expected values mentioned at the end of part A of Risk and Uncertainty include: Actual profit will not be equal to expected value. I understood that it will be equal to one of the profit figures in real life and not the average(expected value)

    We have used expected values for our decision tree calculations. How reliable is our decision tree in real life? Wouldn’t the profit from refurbishing be either one or the other and not an average(expected value)?

  9. avatar says

    great lecture i did benefit a lot …thank you ..but i am wondering why u did not include the aspect of the VALUE OF PERFECT INFORMATION which was recently added to the syllabus?

    • Profile photo of John Moffat says

      If you look at example 1 in the course notes, you will see that perfect information is covered. The changes to the syllabus were announced over six months ago and our course notes were updated immediately.

      Perfect information does not require the use of a decision tree (although obviously you could draw a decision tree just as you could draw one for any decision under uncertainty. In practice you would only draw a tree if it helped you – there is no need in example 1; in the exam you only draw a tree if you are asked to.

  10. Profile photo of cecel says

    Hi John,
    After going through the lecture at least twice, I finally understand every aspect of example 2 that you explain and will attempt the 2012 past paper question. It was a reall………………y good lecture!! I must say though, that like ‘Miss A’ I was wondering how come you included the ‘shut down’ option if Combi did the market research and the result was poor, when the question indicated that they would only be prepared to consider the cheap refurbishment. I felt that you included it because the only reason they would also consider shutting down would be due to cost factor, since if the result is poor it doesn’t make sense to invest in an expensive refurbishing. plus with or without the research the shut down will yield 5m, which is less than the 4.6m return from the cheap refurbishing with the research. Let me know if my logic is heading in the right direction.

Leave a Reply