Performance Evaluating - Call-overwriting program

The most appropriate measure for evaluating the performance of a call-overwriting program is : A. Sharpe Ratio B. Sortino Ratio C. Return to maximum drawdown I don’t know what is call-overwriting program, anyone knows ?

I don’t know the term either. I prefer B. Since call-overwriting program has OPTION figures, so return distribution may have large skewness–>sharpe ratio is not appropriate choice Return to maximum drawdown–>only downside of the return distribution is measured, also not appropriate

B is the correct answer. But any one knows what is call-overwriting program ?

Overwriting means long on a covered put position. In this case, A is clearly not appropriate, since the strategy has a lopsided negative skewness. B. measures negative skewness, so appropriate. I guess C is not appropriate since RoMAD is more used to measure a portfolio from high-water and subsequent low (Hedge fund), thus not 100% appropriate here. It does measure the downside risk for this strategy.

elcfa, Would you please elaborate on the “overwriting (long on a covered put position)” ? “Long a covered put position” means long a put with underlying on hand ?

sorry, covered CALL (not put) = long stock and short call (the first combo learned in the option section)

elcfa Wrote: ------------------------------------------------------- > sorry, covered CALL (not put) = long stock and > short call (the first combo learned in the option > section) Thank you so much !

Sorry, I have 2 questions. 1) Schwester practice exam 3 (P.295) say that if volatility is on the upside; std dev and max drawdown overstate the true risk. How does max drawdown overstate the true risk ? 2) Does downside deviation/ Sortino ratio require normal distribution holds ? If yes, why is it appropriate to measure option like payoff.

  1. I think max drawdown = highest return - lowest return for the period. So, this takes into account the outliers which wouldn’t be reflective of the real returns, thus overstating true risk. (this is my understanding) 2) I think sortino doesn’t need normal distribution assumptions. Isnt it mentioned in the text that sortino is more “appropriate” for hedge funds than sharpe ratio? I would guess this is one of the reaons, and also because of the downside deviation feature, where the manager doesn’t get penalized for volatility below the target risk. Hope i am right with the above explanation.

But the textbook Vol 5, P. 269 also mentioned “departures from normality of returns can raise issues for the Sortino ratio as much as for the Sharpe ratio” so I am wondering why it provides good measure for option like payoff.

Hmm…my bad then…

B_C Wrote: ------------------------------------------------------- > Sorry, I have 2 questions. > > 1) Schwester practice exam 3 (P.295) say that if > volatility is on the upside; std dev and max > drawdown overstate the true risk. How does max > drawdown overstate the true risk ? > Max drawdown not measure highest return – lowest return. It measures (maximum point of return (high water mark) - SUBSEQUENT low point)/high water mark. This is what I understand: Maximum drawdown is normally used in connection with Hedge fund. It is the HF has high kurtosis (fat tail risk) not normally represented in normal equity fund. Therefore, one focuses on the maximum since it happens more often thus must be aware of this risk. In case of Davis’ fund: investing in EMD debts, this kind of fat tails normally does exist (CFAI p 140 v4), thus justifying the use of this measure of his funds’ ABSOLUTE risk-adjust return. However, Schewer seems to ask you to evaluate his performance RELATIVE to Antelope’s criteria because they use a minimum acceptable return. In another word, they are introducing a “misfit” active return. The max drawdown is not relevant since the drawdown comes from the nature of his own benchmark, not his fault since he is following his benchmark while adding his own alpha. In this sense, Sortino is the best since it measures risk RELATIVE to Antelope’s criteria (as measured by MAR), not Davis’ own benchmark. The statement ‘true risk’ must be therefore interpreted as RELATIVE to Antelope’s criteria. > 2) Does downside deviation/ Sortino ratio require > normal distribution holds ? If yes, why is it > appropriate to measure option like payoff. Sortino: does not require normal distribution, but the more the distribution deviates from standard, e.g., more skewed or fat tailed -> the more it is misleading to use Sortino since it does not show those risks clearly (as well as e.g., max drawdown, modified VaR…), In asymmetrical option-like pay off, it is more appropriate to measure than Sharpe, but still not as good as max drawdown, modified VaR, if the pay off in addition has skewness, or high kurtosis. Hope it helps.

Thanks ELCFA, a bit confusing, but I am sure I’ll get it if I read a a few times again. AMC, can you pls. post the official explanation of this question? Thanks

Thanks elcfa. For Q1) I guess 2 explanations for choosing Sortino ratio 1) Schewer ask to evaluate his performance RELATIVE to the MAR. So Sortino ratio is appropriate (as you pointed out). 2) RoMAD uses max drawdown. Max drawdown penalize good performance, treat extremely good performance as risk (similar to sharpe ratio). If high max drawdown is due to extremely good performance. RoMAD will be low. ( do you agree?) For Q2) do you agree on the following ranking ? Positive skewness, high/low kurtosis => ranking: Sortino > RoMAD> Std dev Negative skewness, low kurtosis => ranking: Sortino > RoMAD > Std dev Negative skewness, high kurtosis => ranking: RoMAD> Sortino> std dev Thanks

B_C Wrote: ------------------------------------------------------- > Thanks elcfa. > > For Q1) I guess 2 explanations for choosing > Sortino ratio > > 1) Schewer ask to evaluate his performance > RELATIVE to the MAR. So Sortino ratio is > appropriate (as you pointed out). > Agree > 2) RoMAD uses max drawdown. Max drawdown penalize > good performance, treat extremely good performance > as risk (similar to sharpe ratio). Not sure it is correct RoMAD measures (maximum point of return (high water mark) - SUBSEQUENT low point AFTER that high water mark is reached before the next high water mark)/high water mark So if the performance is a straight line upward with no retrenchment then RoMAD would be very high. In fact, RoMAD of T Bill is infinitive since there is no retrenchment --> denominator =0. If high max > drawdown is due to extremely good performance. > RoMAD will be low. ( do you agree?) > see above. > For Q2) do you agree on the following ranking ? > > Positive skewness, high/low kurtosis => ranking: > Sortino > RoMAD> Std dev > Negative skewness, low kurtosis => ranking: > Sortino > RoMAD > Std dev > Negative skewness, high kurtosis => ranking: > RoMAD> Sortino> std dev > > Thanks No sure what you mean here since they measure different things: If MAR = Risk free then Sortino > Sharpe ratio for positive skewness Sortino < Sharpe ratio for negative skewness

Formula of RoMAD (according to textbook Vol 5, P.268) = average return/ max drawdown or = average return/ largest difference between high water mark and SUBSEQUENT low point Not = high water mark - SUBSEQUENT low point /high water mark Correct? I don’t understand why denominator =0 if the performance is too good, would you please explain?? Thanks

Max Drawdown = high water mark - SUBSEQUENT low point /high water mark Thus RoMAD = average return / [high water mark - SUBSEQUENT low point /high water mark] > > I don’t understand why denominator =0 if the > performance is too good, would you please > explain?? > Take a simple case (no monthly compound, only annual return) to demonstrate the point. Say a HF manager using leverage to drive up his return --> a lot of volatility so start of the year = 1B --> in the middle of the year, his fund is 2B, then it goes into steady decline to bottom out at .6B before ending the year 1.5B. Max Drawdown = (2 - .6)/2 = 1.4/2 = .7 So his RoMAD = (1.5 - 1)/ .7 = .714 so RoMAD does not penalize him for his good return 50% (not the 100% in the middle of the year), but it does penalize him for the downside volatility While another mgr only invest in RF= 5%. The return of RF is a stead line upward with no retrenchment, so max Drawdown = 0 so his RoMAD = 5%/0 = infinity. It goes without saying that it does not make sense to measure RoMAD in this situation. Hope it is clearer

sparty419 Wrote: ------------------------------------------------------- > AMC, can you pls. post the official of > this question? Thanks Sorry, I was on bisiness trip & did not response to you ASAP. The explanation is : The Sortino Ratio considers optionality and does not penalize portfolio manager for volatility derived from outsized positive performance. I don’t know what does it mean by “optionality” here, any one can elaborate ?

For Q 1) Then why did Schweser practice exam vol 1 P. 295 say using max. drawdown as risk measure would penalize Davis because his volatility is derived exclusively from returns in excess of his benchmark? For Q2) Sorry, what I mean here is not numerical ranking, I mean order ranking. Which methods – Sortino, RoMAD, Std dev are preferable under each of the following situations (if the question do not specifically ask to measure performance against MAR)? Positive skewness, high kurtosis Positive skewness, low kurtosis Negative skewness, low kurtosis Negative skewness, high kurtosis

B_C Wrote: ------------------------------------------------------- > For Q 1) Then why did Schweser practice exam vol 1 > P. 295 say using max. drawdown as risk measure > would penalize Davis because his volatility is > derived exclusively from returns in excess of his > benchmark? Not 100% sure about that. If they mean his volatility is exclusively on the the upside (no down side) why does he have such a large max drawdown --> must be coming from benchmark’s own downside volatility --> consistent if what I was saying and the rest of their comment: IN THIS CASE based on the text’s info: must see his ‘true’ risk/performance as defined relative to benchmark or client’s MAR, not because the benchmark in itself is volatile causing large max drawdown. > For Q2) Sorry, what I mean here is not numerical > ranking, I mean order ranking. Which methods – > Sortino, RoMAD, Std dev are preferable under each > of the following situations (if the question do > not specifically ask to measure performance > against MAR)? > Here is what I believe: Whenever you have a departure from normal distribution, SF is no longer preferred. When you have high kurtosis: max drawdown is preferred. When you have skewness: Sortino is preferred to RoMad thus Positive skewness, high kurtosis: RoMAD, Sortino, SF Positive skewness, low kurtosis: Sortino, RoMAD, SF Negative skewness, low kurtosis: Sortino, RoMAD, SF Negative skewness, high kurtosis RoMAD,Sortino, SF