MPS Calculation : L2 Band 10 Failed Failed

slackeranalyst

Does anyone know what approach Institute follow for MPS calculation. Here are two matrix.
First one is based on the correct answers...if you made this many correct answers...the corresponding MPS will be in column 2

Correct Answers MPS
72 60
73 60.83333333
74 61.66666667
75 62.5
76 63.33333333
77 64.16666667
78 65

Second approach is: Institute come up with MPS and then decide, how many correct answers are required to Pass. Here is the catch since correct answers cannot be in decimal...

MPS Correct Answers
50 60
60 72
60.5 72.6
61 73.2
61.5 73.8
62 74.4
62.5 75
63 75.6
63.5 76.2
64 76.8
64.5 77.4
65 78


Please scratch your head and think if you have 73 correct answers and MPS is 60.84...what decision Institute will make...there can be many other cases like this....

Guys, I've compiled some 50+ result matrix (only band 10 and barely Passed cases) with all probabilities and can conclude the MPS for this year L2 is in this close range of 60.5 to 61.5.

Rasec

60.5 to 61.5 cannot be right - this would mean the effective pass rate was approx 40% (guess 1 in 3 of the remaining). This is far too low.

Windjammer

RETABULACTION TIME!

bchadwick

My guess is if you spent as much time trying to figure out the MPS as you would have studying for L2 you would have passed it.

Not trying to be mean but who cares what the MPS is? Do better next time.

SkipE99

I think that there might be a problem calculating the bands. First of all, we can't assume that it is distributed equally - kurtosis, or left/right tail bias. What would be great is to obtain from all the samples the absolute worst set of matrices that have a pass - then, (and this is the difficult part) assume the highest values possible. Only then would we be able to eliminate all possibilities of a higher grade.

Since the results that show >70 can really set our sample to 100... we can modify this to 90. Then, we could get a sample of 5-10 and make some assumptions.

Or we can have a hunch that the MPS was around 63... and let whoever takes L2 next year worry about this.

CPATrader

60.5 to 61.5 is in terms of percentage...so the required correct answers must be in the range of 72 to 74.

The Band 10 fail will be in very narrow bracket...my basis is I've also taken into account few Band 9 and 8 samples to make sure the bias factor for Band 10 to be kept at minimum.
As per me if you have to see MPS as minimum correct answers out of 120...the band 10 range will be 69 to 73. Below that another 4 will be Band 9 and so on.

The confidence level is 95% as rigtly said

I've got my retabulation result in just 3 days with no change as expected only to waste another 100 F$$$.

Swanand

At the end of the day you're still guessing what people scored in some of the topics and making several large assumptions. The larger the weighting, i.e. FRA & Equities, the more you end up guessing. Also, even though you are just sampling the Band 10 fails even that group has its own score distribution. Some of the Band 10s were presumably one question away from passing while others were maybe 4 or 5.

I can't imagine that the CFAI would set the bar that low. Plus, we really don't know how much of an impact the Ethics adjustments has on passing or failing. The CFAI will never divulge the secrets of the MPS so it's not really worth wasting time over it.

bleach

That's interesting. Seems low to me. Do it for last years level 3 results. lol



Edited 1 time(s). Last edit at Wednesday, August 3, 2011 at 11:55AM by JP_RL_CFA.

MiniMe7

60.5 to 61.5 eh? would you say that's your 95% confidence interval?

SpyAli

I think CFAI only defines the MPS in terms of number of correct answers, which means that only whole numbers are used.



Edited 1 time(s). Last edit at Wednesday, August 3, 2011 at 07:14AM by cityboy.