Nice… 0 for freakin’ 3. Good set. But I’d still argue on number 1. You could have two portfolios – one with an extremely low probability of a very large loss, and one without, that still had an identical VAR measure. However, that distinction (which is, after all, information) would be lost if you used the VAR measure. The VAR does NOT use all available information. It ignores the magnitude of losses beyond the sensitivity threshold.
Absolutely. ^
Correct me if I’m wrong but VaR does use all the available DATA. To get a VaR measure, you take an entire distribution of data, make a nice little histogram and measure a particular chuck of it. You don’t do this with semivar. The calculation excludes data below a certain threshold. Look at it this way, if you took a histogram of the distribution of returns above a semivar threshold, you would have a very different looking graph than if you included the entire distro.
I do not know whether Jack collected these questions, but; for 1st one: the key is “He requires a measure that will use all the data so no information will be lost”. This is called a sufficient statistic in economics or/and statistics. neither semi-var and var qualify. for 2nd one: it is an empirical question. Rentech charges 5% annual fee and 20% performance fee, but as long as return is good who cares? I bet they have very high info ratio as well. for the 3rd one: "“A proxy hedge only makes sense when we have high correlations for the currency movements.” it is of course true. If there is no high correlation, it is not called proxy hedge, it will be called statistical arbitrage. these questions are really bad questions.
I just got these questions out of the qbank from Schweser… The questions that I post are usually ones that I got wrong or found tricky. This doesn’t necessarily mean they’re good questions. Infact if the general consensus is that the questions are bad then it makes me feel a little better as then I can blame the question (“if only” defense?). Feel free to ignore my threads in the future… CFAMonster Wrote: ------------------------------------------------------- > I do not know whether Jack collected these > questions, but; > > for 1st one: the key is “He requires a measure > that will use all the data so no information will > be lost”. This is called a sufficient statistic in > economics or/and statistics. neither semi-var and > var qualify. > > for 2nd one: it is an empirical question. Rentech > charges 5% annual fee and 20% performance fee, but > as long as return is good who cares? I bet they > have very high info ratio as well. > > for the 3rd one: "“A proxy hedge only makes sense > when we have high correlations for the currency > movements.” it is of course true. If there is no > high correlation, it is not called proxy hedge, it > will be called statistical arbitrage. > > these questions are really bad questions.
CFAMonster Wrote: ------------------------------------------------------- > I do not know whether Jack collected these > questions, but; > > for 1st one: the key is “He requires a measure > that will use all the data so no information will > be lost”. This is called a sufficient statistic in > economics or/and statistics. neither semi-var and > var qualify. I keep harping on this but VaR uses all the data; no information is lost. I don’t understand why this is so hard to get.
recycler, you think no information is lost because you use all the information in the INPUT others think information is lost because the OUTPUT doesn’t give you all the information.
Name one output that gives you all the information? There isn’t one. Variance doesn’t tell you anything about skewness or tail fatness. I think the for the exam, what we need to understand is that no single measure gives you a complete picture of risk and that semi-var excludes portions of the distribution in it’s calculation. I hate this quant crap. Gets me all worked up. Plus I just fucking hate Schweser questions.
CFAMonster Wrote: ------------------------------------------------------- > I do not know whether Jack collected these > questions, but; > > for 1st one: the key is “He requires a measure > that will use all the data so no information will > be lost”. This is called a sufficient statistic in > economics or/and statistics. neither semi-var and > var qualify. > > for 2nd one: it is an empirical question. Rentech > charges 5% annual fee and 20% performance fee, but > as long as return is good who cares? I bet they > have very high info ratio as well. > > for the 3rd one: "“A proxy hedge only makes sense > when we have high correlations for the currency > movements.” it is of course true. If there is no > high correlation, it is not called proxy hedge, it > will be called statistical arbitrage. > > these questions are really bad questions. 100% agree.
recycler Wrote: ------------------------------------------------------- > Name one output that gives you all the > information? There isn’t one. Variance doesn’t > tell you anything about skewness or tail fatness. > Normal distribution as described by mean and S.D. Tells you EVERYTHING about the distribution.
CFAMonster Wrote: ------------------------------------------------------- > I do not know whether Jack collected these > questions, but; > > for 1st one: the key is “He requires a measure > that will use all the data so no information will > be lost”. This is called a sufficient statistic in > economics or/and statistics. neither semi-var and > var qualify. > > for 2nd one: it is an empirical question. Rentech > charges 5% annual fee and 20% performance fee, but > as long as return is good who cares? I bet they > have very high info ratio as well. > > for the 3rd one: "“A proxy hedge only makes sense > when we have high correlations for the currency > movements.” it is of course true. If there is no > high correlation, it is not called proxy hedge, it > will be called statistical arbitrage. > > these questions are really bad questions. That is why I only do Schweser Video and notes BUT not questions bank because they can be highly subjectives and incorrect. ( Here, i am talking for the level 3 only)
VaR won’t let you know the expected loss in case a rare event happens. In fact, in level I, Level II and even in GRE, anytime you see “all”, you eliminate that choice.
If you know it is normal, why do you need VaR? mwvt9 Wrote: ------------------------------------------------------- > recycler Wrote: > -------------------------------------------------- > ----- > > Name one output that gives you all the > > information? There isn’t one. Variance doesn’t > > tell you anything about skewness or tail > fatness. > > > > Normal distribution as described by mean and S.D. > Tells you EVERYTHING about the distribution.