Hi ajsa,
they are saying they first find the 5th %ile score (sort of like VaR but in the + direction!); then they take a ratio of that; e.g., 5th %ile = 93% correct, then passing is P*95%. But they calibrate (P) after .... Therefore, I say, it isn't useful ex ante ... just a variation on "grading on a curve"
Re 1.65 or 1.645? definitely should not matter, 1.65 should be sufficient
Re: compounding: as you know from the questions, they really *need* to say. But rule of thumbs: if bond, "default" to semi-annual (Tuckman), otherwise continous is a fine default (but questions should say....)
David
they are saying they first find the 5th %ile score (sort of like VaR but in the + direction!); then they take a ratio of that; e.g., 5th %ile = 93% correct, then passing is P*95%. But they calibrate (P) after .... Therefore, I say, it isn't useful ex ante ... just a variation on "grading on a curve"
Re 1.65 or 1.645? definitely should not matter, 1.65 should be sufficient
Re: compounding: as you know from the questions, they really *need* to say. But rule of thumbs: if bond, "default" to semi-annual (Tuckman), otherwise continous is a fine default (but questions should say....)
David
