How to guess on CFA

Borlta wrote:I guess I wasn’t clear enough. You’re right that you should always, before doing anything else in EPS questions, determine whether you’re dealing with dilutive/antidilutive securities. What I was trying to say is that if you KNOW the securities are dilutive but are not quite sure how to calculate the diluted EPS, you should calculate normal EPS just to see if it’s an option. In my experience doing mocks and QBank questions, it often is. If it is, eliminate it and, if you’ve spent a lot of time on the question and aren’t getting anywhere, guess between the other two answers. Again, you double your chances of guessing the correct answer if you can eliminate one option.
Click to expand...
Sorry Borlta, but you’re incorrect. This isn’t a “Monty Hall problem” (most recently famous from the movie 21).
http://en.wikipedia.org/wiki/Monty_Hall_problem
If you can eliminate one 1 answer, you have 50/50 on the other two. Please read link above if have any questions
 
The way i interpret it, it’s a Monty Hall problem. If you guess randomly, you have a 1/3 chance of getting the question right. If you can for certain eliminate one answer, that 1/3 goes into your corner. The odds of getting one of the other two answers correct are still 1/3, so the odds of guessing between the two becomes 1/3 plus 1/3, or 2/3.
EDIT: Will read up on the link at a later juncture. I may have misinterpreted the problem. Monty Hall problem or not, though, eliminating obviously false answers is always a good idea and increases your odds of guessing a correct answer.
 
Monty Hall problem applies when you are shown a DIFFERENCE answer is incorrect to the one you’ve chosen. I..e you think it’s A, and then you are told it is NOT B. Thus, you should switch your answer to C. Does not apply in this case at all.
Anyway, the EPS example doesn’t really apply because the premise is you’ve got 10 questions remaining and just under 45 seconds left in the test.
I think you can apply a bit of logic to the answer patterns (not sequencing but the answers within each question) and imporve your odds. If I had time, I’d go through an exam using this logic to see what I would score. But I have not the time.
 
SimplyAdvised wrote:
Here’s how you guess when you have NOOO idea and no time to really read the question:
Q1) blah blah blah blah
a) 7.31
b) 7.56
c) 24.34
(guess a or b)
Q2) blah blah something goes up, something goes down blah blah
a) higher
b) no change
c) lower
(guess a or c)
Q3) This question is written in ancient sumarian
a) -50
b) - 10
c) +50
(guess A - answer is probably negative since two of the answers are negative, and it’s probably “50” since two of the answers are 50)
Click to expand...
Worst advice ever
 
I consider myself an expert guesser:
If two time periods are involved and you need a calculation - Do what you think you need to do and if your calculation isn’t one of the choices, take the first difference of all measures and see if anything pops out that would make your calculation fit one of the given choices. If it makes logical sense then that’s a good guess.
If its a general knowledge type question - Look for terminology you know is real as there are frequently fake terms in the wrong answers (or correct answers if it is a “least correct” question - you know what I mean).
Try to roll up sub components into their super-components and see if it helps with the logic of the question.
If you can’t arrive at a precisely correct answer, choose the one you’re closest too.
If you have 0 ideas, don’t choose the outlier.
When I’m reading through questions that don’t require a calculation, I always cross out answers that I am sure are not correct. I also read all answers regardless of whether or not I think I’ve already found the correct one.
Also always look for mutually exclusive answers. They pop up from time to time. I remember one where the choices for a least correct problem were, “has no impact on cash flow from operations”, “impacts cash flow from operations this way”, “impacts cash flow from operations this other way”. The only answer was the first answer because if the first statement were true then both the second answers would have to be false which would give two possible answers.
If you’re out of time and just have to fill in something, you’re best bet is to choose the same answer for every question. I think this is the most statisically effecient method, assuming the distribution of the answers is random. Anyone with a good stats background confirm/reject this?
 
“If you go straight to your scantron and just fill in the the remaining 10 without looking at the answers, you’ll have a 1/3 chance on each. No difference between AAAAAAAAAA or ABCABCABCA. “
This was asked in our Quant lecture. According to the instructor, who is a former math teacher, your advice is wrong.
If guessing, you only want to guess one letter for all the answers.
Otherwise you are radomizing the radomizer, which decreases your odds significantly.
 
Broker11 wrote:
“If you go straight to your scantron and just fill in the the remaining 10 without looking at the answers, you’ll have a 1/3 chance on each. No difference between AAAAAAAAAA or ABCABCABCA. “
This was asked in our Quant lecture. According to the instructor, who is a former math teacher, your advice is wrong.
If guessing, you only want to guess one letter for all the answers.
Otherwise you are radomizing the radomizer, which decreases your odds significantly.
Click to expand...
I don’t think your instructor is right. I also don’t understand what you mean by randomizing the randomizer, and why randomness squared would decrease the odds significantly (or at all). I may be missing something key, but I can’t see why this would make sense.
The expected “return” of any single answer will always be 1/3, unless you’re assuming some form of bias in the CFA exam. The probabilities don’t know which letter you’re talking about, which color is your marble ball or whatever. Remember that those are independent events.
So the expected return on any number of answers will be n/3. If you’re completely guessing the whole exam your expected return is 33.3%.
If you assume CFAI wants an equal or almost equal number of answers to be either A, B or C then choosing a single letter may reduce the expected volatility of the results, especially if you have a big sample. However, especially with a big sample (as in you’re completely lost on most of the exam ), reducing volatilty may be a bad strategy - if you completely guess 50% of the test with zero volatility you got 16.7% - you probably fluked the exam already (your max possible score becomes 66.7%).
Bottom line - I don’t think you should game the exam, especially like that. The only benefit I see on guessing always A or whatever is saving time (once you’re lost, you won’t waste time making a decision).
 
Broker11 wrote:
“If you go straight to your scantron and just fill in the the remaining 10 without looking at the answers, you’ll have a 1/3 chance on each. No difference between AAAAAAAAAA or ABCABCABCA. “
This was asked in our Quant lecture. According to the instructor, who is a former math teacher, your advice is wrong.
If guessing, you only want to guess one letter for all the answers.
Otherwise you are radomizing the radomizer, which decreases your odds significantly.
Click to expand...
Your ‘former math teacher’ is entirely incorrect. “Randomizing the randomizer” is a meaningless phrase. None of the results are “partial random” (a phrase which verges on being an oxymoron), so you can’t make it any more random.
The important assumption here is that the correct letter answers are independent of each other. If this is the case (most standardized tests are made this way), then we the Binomial Distribution model is correct here. independence means the outcome of one answer doesn’t the next. ie if the first answer is A, the next is equally likely to be A,B or C. As long as each answer has equal probability, it doesn’t matter which you choose.
Regardless of ABCABCABC or AAAAAAAAA, your e(x)=np
 
You must log in or register to reply here.
Back
Top