60 questions per session, 120 questions in total. assumption 1: passing score is around 70% assumption 2: choosing the correct answer out of the 3 possible choices follows standard normal distribution. assumption 3: candidates have been focusing on Ethics, FRA and Equities during their studies and will do slightly better in these sections than other. During the exam, the minimum number of questions that we need to do correctly is 68/120 questions. This leaves a room of 54 questions that we are unsure or have no clue about. If we guess them all, we will, on average, get 18 questions right. thats 18 extra marks. (18+68)/120=70%. lol, if u really think about, getting a minimum of 56.7% (68/120) of the questions correct and let probability do the rest might just pass you! Alright~~Giggaddyy Edited 2 time(s). Last edit at Saturday, February 26, 2011 at 05:51PM by passme.
typo: 120-68=52, not 54 52x 0.3333 = 17, not 18
I like it. however you are assuming that all the ones you “know” you will get right ie not misread the ? or get confused by the answer choices. still a cool calculation
like you said its obvious lol if you got to level II and you cant do such math, boy you defied some odds… but anyway, i guess it all goes to how risk averse you are…do you realy wana gamble on it there is a very good chance you get less than 18 right… with such a small number of runs, your avg wont be at 0.33333 for sure… i am sure you can calculate the odds of that …
the sum of probabilities for outcomes less than 18 of 54 correct ie for 17, 16, 15…0 correct is 0.447170018 so assuming you somehow managed to know for 68 correct and you randomly guesses the rest, you chances of passing are only 55.28% i just though i would share this in case so crazy guy decides to follow this, i am sure thats no what you ment for though 
also will someone please help me on this "choosing the correct answer out of the 3 possible choices follows standard normal distribution. " each run follows of a binomial distribution with p success=0.33333 p fail=0.666666666 so nothing “standard normal about it” and the distribution of the number of successes in 54 runs is not standard normal either because you have a chance of success diff from fail so not normal distribution… i am not trying to punk the guy, just wana see if my level 1 skills are still okay
bah, ill simplify it for everyone here: assuming CFAI won’t fail anyone who makes 70% on the exam, you need 84/120 questions right. -------- If you can guarantee 67 questions correct, that means you guessed on the remaining 53 questions. Out of these 53 guesses, you have a 1/3 chance on guessing correct. This means statistically speaking: you guessed 17.66 questions right. So, your 67 guaranteed + 17.66 guessed right = 84.66 questions right ---- So, THEREFORE, your guaranteed 67/120 means you only need 56% on the test. Your free 17.66 questions should push you into 70% overall.
ugh, why is it that you guys dont get this you wont get 17.666 of the 53 correct… thats just the mean value, there is a good chance yo get less than that, even if he chance is 1/3 if you roll the dice six time, are you garauneed to get a “6” once ?
ya, i am just bored and put this out so that maybe this can boost some confidence. ahmadmadoff thx for ur input, lol way overboard brother…
ahmadmadoff Wrote: ------------------------------------------------------- > also will someone please help me on this > > "choosing the correct answer out of the 3 possible > choices follows standard normal distribution. " > > each run follows of a binomial distribution with > > p success=0.33333 > p fail=0.666666666 > > so nothing “standard normal about it” > > and the distribution of the number of successes in > 54 runs is not standard normal either because you > have a chance of success diff from fail so not > normal distribution… > > i am not trying to punk the guy, just wana see if > my level 1 skills are still okay this is pretty much it
passme Wrote: ------------------------------------------------------- > ya, i am just bored and put this out so that maybe > this can boost some confidence. ahmadmadoff thx > for ur input, lol way overboard brother… bro, the code bellow will tell you how overboard i went, i ran a simulation of one billion test takers at one point lol… i just needed something to get me away from finance and back my computer roots gona hit he sack, study hard, i am gona add a question to his thread while we are here, see it next Dim correct As Decimal = 0.33333336666 Dim wrrong As Decimal = 0.66666666666 Dim ran As New Random Dim results(54) As Integer Dim tries As Integer = 10000000 Dim correctFromRuns As Integer = 0 While tries > 0 Dim x As Integer = 54 correctFromRuns = 0 While x > 0 If ran.NextDouble() < 0.3333333 Then correctFromRuns = correctFromRuns + 1 End If x = x - 1 End While tries = tries - 1 results(correctFromRuns) = results(correctFromRuns) + 1 End While Dim y = 0 While y < 55 TextBox1.Text = TextBox1.Text + vbCrLf + (results(y) / 10000000).ToString y = y + 1 End While End Sub
when i run the simulation these are the results ( you get the same with binomial and combination formula) 0 0 3E-07 1.4E-06 5.5E-06 2.91E-05 0.0001299 0.0004234 0.0012679 0.003201 0.0072231 0.0144879 0.0259056 0.0419101 0.0614195 0.0818618 0.0998313 0.1115398 0.1144466 0.1083755 0.0949009 0.0769714 0.0575343 0.0400749 0.0259469 0.0155324 0.0087066 0.0044981 0.0021706 0.0009624 0.0003991 0.0001602 5.67E-05 1.85E-05 5.7E-06 1.2E-06 4E-07 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 does anyone know the reason for having more chance of getting more than the mean of 18 rather than bellow it, given that the probability of geting wrrong is bigger than getting right? i meen obviously this is correct, but my first instinct was telling me you are more likely to get less than 18…
ahmadmadoff - go study instead of this.
Palantir you and my guilty concious are telling me this… i promised my guilty concious i pick up the pace tommorw, since its 3am now… what can i do for you lol… passme started it for fun, and i joined for fun, but i just take anything i do in life seriously, even fun… you can imagine, hanging out with me must be realy fun
nice job ahmad if i had the coding skills i would have done the same 
Enjoy your saturday nights brothers. 
ahmadmadoff - what do you mean, the probability for getting above 18 is higher. Are you talking about the sum of the cumulative probabilities from 0-17 vs 19-54? what language was your code in anyway? The binomial distribution does have some amount of skew and kurtosis beyond the normal distribution for p != 1/2, though it should drop off quickly as N grows. But even here where N=54 the normal distribution gives close-enough answers (ie, probability to get ‘k’ right w/ p=1/3 and N=54) : 1/sqrt(2*pi*N*p*(1-p)) * exp(-(k-18)^2/(2*N*p*(1-p)))
This again, I brought this up about level 1, http://www.analystforum.com/phorums/read.php?11,1148045,1148045#msg-1148045 but ahmad is right, of course you are not guaranteed to get 17+ right, I think we all know that, but this was just for fun. At least it made me feel more confident last year BTW, why don’t all the nerds on this forum team up and make some quant portfolio models, we can sell them!!
> BTW, why don’t all the nerds on this forum team up > and make some quant portfolio models, we can sell > them!! I dono man, if we were at that level of “nerdiness” we probably would not be here in the first place… Probably already have such jobs and would not even need the CFA to prove anything… Just a thought…
Well, that would mean, all nerdy portfolio/fund managers would not have their CFA charter. Even Einstein went and got his PhD