Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
The Owl honor roll pdf is highly troubling to me. Any such listing whose criteria somehow leave off JABAX, GLRBX, various Intrepid funds including ICMBX, GABSX, FLPSX, all Yacktman products, PRBLX, while listing Weitz and D&C confuses me utterly. I believe I get why ARLSX is not present (time). And yes overall, I do see what the criteria are, and I have said in the past that I simply must study all of these data harder; so I'm going to redouble my efforts in these regards. Cuz I have to. Still, a puzzling honor roll for my brain.
Hasty example: To see the FFNOX / FAMRX value supposedly added by Dierdorf, Stein, Sharpe et alia over 1/3/4/5y, just compare their work with, say, AOA and AOR, in addition to the balanced vehicles given above.
Reply to @davidrmoran: Ha! Hi sir. The Honor Roll stuff is easiest of all. Simply top absolute returning funds in their respective categories during past 1, 3, and 5 year periods.
Without digging deeper, the omissions you mention tend to be top risk adjusted returning funds in their categories, so they are more likely to end up on the Great Owl list than the resurrected Honor Roll list.
And Three Alarm Funds are the antithesis of Honor Roll, simply bottom absolute returners in their categories during past 1, 3, and 5 year periods.
Strictly determined by monthly total return numbers. Strictly quantitative. No qualitative assessments made with any of these lists.
Hope that helps.
And, hope all is well.
PS. Looking to publish the piece on persistence we have discussed in David's next commentary or two.
PS.PS. AOA and AOR were just shy of 5 years old as of end of 3rd quarter, so they were not included in either the Honor Roll or Three Alarm rankings. But I believe they will be included in 4th quarter update.
Reply to @Charles: Yes, okay, as I thought, and I see the benefits of going back only 5 years, rather than 5.5 or 6. From your honors one might conclude that D&C and Weitz had more downside protection than Parnassus or the Yacktmans; after all, you do give a column for risk rating even in this quant listing (why?). Still, a quick check fails to see how GLRBX did not make the cut when it outperforms SWCGX significantly. So again, assuming I am comparing oranges properly, I am missing something.
OK, here are the numbers for GLRBX in the Three Alarm/Honor Roll system:
GLRBX was really close, but fell just shy in the 5 year period from the other top funds in the conservative allocation category. Note that SWCGX is not an Honoree either, I believe it was just shown as a reference/benchmark fund.
The risk rating is pure Roy Weitz, resurrected. It signifies relative risk within the category. So, in this case, it says GLRBX is riskier than SWCGX . The assessment uses standard deviation during last 3 years.
>> signifies relative risk within the category. So, in this case, it says GLRBX is riskier than SWCGX. The assessment uses standard deviation during last 3 years.
Hmm and huh. I see.
I think that conclusion is easily arguable (was going to write 'nuts') and would suggest somehow modulating this criterion. Or omit relative risk within the category.
Maybe you will address the smoothing of resurrections in your persistence crunching.
Easy to say when I am not doing the work, to advise you to omit the column or refine it. Many funds look superduper post-08. I think a system that somehow quantitatively alerts MFOers to the dip performance (at least two dips, say, including last March or whenever it was) of Weitz, Berkowitz, and D&C compared with (say) the Parnassus, Yacktman, Gabelli, and AMANX crowd (I did not even mention the last guy) would be useful. You know how people here chase (3y) performance.
The same argument could be made if you crunch the graphs of the other balanced funds I mentioned, ICMBX and JABAX, vs their peers. In other words, there is something slightly off with including those two (good) Fido funds and not including those two dip-protecting ones. And yes, I realize the M* categories may be different.
And again, it is easiest (repeatable, justifiable) simply to rely on the cut-and-dried math, I know. Easier not to take the pains to devise a dip column to include for honoring, which you have elsewhere but not here. I'm just trying (flailing) to find something rather more helpful to MFOers than this seemingly oddly exclusionary pdf, all while not being the one who does the labor and is not the stat comparer.
Well David and his crew have done a super job this year with content and enhancements. I sincerely hope all you commenter’s also thank David by making a contribution financially as this site doesn't run for free. Remember, you get what you pay for and for and don't kill the golden goose.
Your recent exchanges serve as terrific examples of the difficulties embodied in an individuals risk definition and risk measurement standard. It is surely a multifaceted issue that is heavily influenced by personal considerations. I suspect we all harbor slightly different sets of risk measures. I find nothing disquieting about that observation. To each his own.
Perhaps my memory is faulty (a reasonably high likelihood), but I seem to recall that the risk measure in the old FundAlarm series included the 3-year average return in its definition. It sort of assumed a Normal returns distribution and calculated the average 3-year return minus two times the 3-year standard deviation. Since it was based on monthly values, it incorporated 36 data points into each calculation. I acknowledge that I might easily be in error on this matter.
If accurate, that formulation identifies the 2.5 % likelihood that the next return will be below the computed average value, which may be negative. I do not find this calculation to be an inspired risk measure. I’m not much interested in an outcome that has only a 2.5 % chance of happening.
Using the same 3-year average return and standard deviation, I prefer to calculate the probability of a zero return year. That’s something an investor really fears. The evaluation is easily accomplished by referring to a common “Table of the Standard Cumulative Normal Distribution” listing found in most statistical handbooks.
I simply ask the following question: How many negative standard deviations are required to force the annual 3-year return down to the zero threshold? That number defines the probability of a year’s zero return reward. That probability merits attention.
To illustrate, say that the equity market had a 10 % annual 3-year average with a 15 % standard deviation. That means that only a negative two-thirds standard deviation event will drawdown my anticipated return to the zero level. That’s a meaningful target number. What is the probability of such a happening?
Using the standard normal distribution tables, a one-sided probability of exceeding that level is 25 % (accuracy beyond two places is sheer nonsense). Hence if the quoted values reflect my portfolio, I would anticipate a negative annual reward approximately 25 % of the time. That’s reasonably representative of the S&P 500 actual performance record.
Now that estimate is meaningful to an investor. He must be prepared to accept and/or reject and minimize that probability. Certainly one easy option is to design a portfolio with a much lower standard deviation expectation without sacrificing annual returns too much (free lunches are rare, but one exists in this arena). Portfolio diversification offers opportunities in this realm.
As another measure of risk, I would also be interested in the historical annual maximum drawdown that a portfolio, similar to my current design, experienced. This is one of many alternate measures of risk. You get to pick your own poison, your own risk measurement standards. A magic elixir doesn’t exist.
I hope this post contributed to the general and genial discussion. We are definitely not in a debate mode, at least not yet.
On a different note,today is Giving Tuesday.Please consider those not as fortunate as we may be.I will be giving to The Salvation Army,Partners in Health and Doctor's Without Borders.
Reply to @MJG: Hi sir. You are of course correct, although Mr. Weitz only deducted one sigma:
Like in the original Three Alarm list, a fund’s Risk Rating is assigned based a “potential bad year” relative to other funds in the same category. A Risk Rating of “2” (highlighted in red) goes to the highest risk funds, while “-2” (blue) goes to the lowest risk funds.
“Risk” in this case is based on the 3 year standard deviation and return values. Specifically, two standard deviations are subtracted from the return value. The result is then compared with other funds in the category to assign a rating. The rating is a little more sensitive to downside than the original measure as investors have experienced two 50% drawdowns since the Three Alarm system was first published.
Hope that better explains the result.
As for the healthy debate, fortunately, that never ends.
Break, break.
Markets have stalled a bit lately. Awaiting next Fed move, direction of 10 year Treasury, and holiday spending, etc. And, surely, happy with the healthy advance this year...perhaps some folks taking money off table, or simply not jumping-in on the dip.
Reply to @Charles: Well, this is most interesting, and thanks for bearing with repeats of your method and criteria. I would suggest perhaps thinking further about recency and period span, among other things. Look for example at the 08-09 dip performance of BRUFX and JILMX vs GLRBX and the seemingly splendid WHGIX, new to me, and then check your risk ratings for the four of them. (Or *is* WHGIX splendid? Check out the M* stars it receives, interesting it's so low, and also note its M* categorization different from yours.) Then also check their dips 9-10/11. Without knowing the 6-7y performance, an MFO reader of your honor table could be led into drawing misleading conclusions about volatility, and much else, for these two inclusions. (It's clear that more recently they're on a tear, so maybe that's all that needs to be said.)
Sorry for this late additional commentary. Since the originating postings highlighted the risk aspects of the investment process, my initial contribution also focused on that side of the investment equation.
When responding I solely emphasized the risk part of the conventional risk/reward tradeoff. I completely failed to document the detrimental impact that price volatility plays on net actual annual returns. We all understand that, because of the iron rules of mathematics, it takes a higher positive return percentage to recover from an unlucky negative return percentage.
A simple algebraic equation captures the difference between an average annual return and the actual annual return after correcting for the negative influence of a portfolio’s volatility. I apologize to those folks who hate equations, but, by necessity, here it is: actual annual return rate equals the average annual return rate minus one-half times the square of the annual standard deviation rate.
So portfolio or individual holding volatility (as measured by standard deviation) always detracts from the average annual returns. It is forever a drag on returns. For a portfolio that combines many investment components, the drag magnitude also depends on the correlation coefficients between all the individual elements.
Correlation coefficient is dynamic and changes over time and is dependent on market conditions. But it never gets to perfect correlation (absolute co-movement), and therefore always benefits a portfolio. When markets are challenged, these dynamic correlations do tend to gravitate towards the “perfect” level, thus attenuating its beneficial aspects. But they never completely disappear. That’s why diversification is such a powerful risk mitigation strategy; the tactic is indeed nearly a free lunch.
Let’s complete the picture by calculating the impact of volatility on the example that I provided in my earlier posting on the topic. With an annual return of 10 % and a portfolio standard deviation of 15 %, the portfolio’s actual annual return is 0.10 – (0.5 X 0.15 X 0.15) or 0.08875. Therefore, the actual annual return rate is degraded to 8.88 %. Over any period of time, the actual annual return rate is used when calculating accumulated wealth.
The investment goal should be to keep actual annual return rate at as high a level as your risk tolerance and practical return prospects permit.
High portfolio returns volatility operates negatively in both the returns dimension and in the risk dimension. High portfolio volatility erodes end wealth over time. High portfolio volatility increases the frequency likelihood of a zero annual return over a portfolio’s lifespan.
So work and monitoring is needed to minimize a portfolio’s overall standard deviation metric. From a practical perspective, cutting a portfolio’s standard deviation by half is an achievable target with only a small sacrifice in average annual return rate. Some of that sacrifice will be partially recovered by reducing a portfolio’s overall volatility.
Enough. Again, sorry that in my haste, I omitted a discussion of the influence that volatility has on portfolio performance. Thank you for your patience.
Problem with this thesis is that it is not possible to predict what diversification decided today will lead to reduced or increased volatility in the future and what will or not lead to better performance. There is no "ideal" diversification except by hindsight.
Comments
http://www.valuewalk.com/2013/12/hottest-links-high-noon-for-contrarians-roll-call-of-the-wretched/
The Owl honor roll pdf is highly troubling to me. Any such listing whose criteria somehow leave off JABAX, GLRBX, various Intrepid funds including ICMBX, GABSX, FLPSX, all Yacktman products, PRBLX, while listing Weitz and D&C confuses me utterly. I believe I get why ARLSX is not present (time). And yes overall, I do see what the criteria are, and I have said in the past that I simply must study all of these data harder; so I'm going to redouble my efforts in these regards. Cuz I have to. Still, a puzzling honor roll for my brain.
Hasty example: To see the FFNOX / FAMRX value supposedly added by Dierdorf, Stein, Sharpe et alia over 1/3/4/5y, just compare their work with, say, AOA and AOR, in addition to the balanced vehicles given above.
Does that make us part of the Twittersphere?
Without digging deeper, the omissions you mention tend to be top risk adjusted returning funds in their categories, so they are more likely to end up on the Great Owl list than the resurrected Honor Roll list.
And Three Alarm Funds are the antithesis of Honor Roll, simply bottom absolute returners in their categories during past 1, 3, and 5 year periods.
Strictly determined by monthly total return numbers. Strictly quantitative. No qualitative assessments made with any of these lists.
Hope that helps.
And, hope all is well.
PS. Looking to publish the piece on persistence we have discussed in David's next commentary or two.
PS.PS. AOA and AOR were just shy of 5 years old as of end of 3rd quarter, so they were not included in either the Honor Roll or Three Alarm rankings. But I believe they will be included in 4th quarter update.
OK, here are the numbers for GLRBX in the Three Alarm/Honor Roll system:
GLRBX was really close, but fell just shy in the 5 year period from the other top funds in the conservative allocation category. Note that SWCGX is not an Honoree either, I believe it was just shown as a reference/benchmark fund.
The risk rating is pure Roy Weitz, resurrected. It signifies relative risk within the category. So, in this case, it says GLRBX is riskier than SWCGX . The assessment uses standard deviation during last 3 years.
Got it?
k; I erred in using SWCGX when not honored.
>> signifies relative risk within the category. So, in this case, it says GLRBX is riskier than SWCGX. The assessment uses standard deviation during last 3 years.
Hmm and huh. I see.
I think that conclusion is easily arguable (was going to write 'nuts') and would suggest somehow modulating this criterion. Or omit relative risk within the category.
Maybe you will address the smoothing of resurrections in your persistence crunching.
Easy to say when I am not doing the work, to advise you to omit the column or refine it. Many funds look superduper post-08. I think a system that somehow quantitatively alerts MFOers to the dip performance (at least two dips, say, including last March or whenever it was) of Weitz, Berkowitz, and D&C compared with (say) the Parnassus, Yacktman, Gabelli, and AMANX crowd (I did not even mention the last guy) would be useful. You know how people here chase (3y) performance.
The same argument could be made if you crunch the graphs of the other balanced funds I mentioned, ICMBX and JABAX, vs their peers. In other words, there is something slightly off with including those two (good) Fido funds and not including those two dip-protecting ones. And yes, I realize the M* categories may be different.
And again, it is easiest (repeatable, justifiable) simply to rely on the cut-and-dried math, I know. Easier not to take the pains to devise a dip column to include for honoring, which you have elsewhere but not here. I'm just trying (flailing) to find something rather more helpful to MFOers than this seemingly oddly exclusionary pdf, all while not being the one who does the labor and is not the stat comparer.
Your recent exchanges serve as terrific examples of the difficulties embodied in an individuals risk definition and risk measurement standard. It is surely a multifaceted issue that is heavily influenced by personal considerations. I suspect we all harbor slightly different sets of risk measures. I find nothing disquieting about that observation. To each his own.
Perhaps my memory is faulty (a reasonably high likelihood), but I seem to recall that the risk measure in the old FundAlarm series included the 3-year average return in its definition. It sort of assumed a Normal returns distribution and calculated the average 3-year return minus two times the 3-year standard deviation. Since it was based on monthly values, it incorporated 36 data points into each calculation. I acknowledge that I might easily be in error on this matter.
If accurate, that formulation identifies the 2.5 % likelihood that the next return will be below the computed average value, which may be negative. I do not find this calculation to be an inspired risk measure. I’m not much interested in an outcome that has only a 2.5 % chance of happening.
Using the same 3-year average return and standard deviation, I prefer to calculate the probability of a zero return year. That’s something an investor really fears. The evaluation is easily accomplished by referring to a common “Table of the Standard Cumulative Normal Distribution” listing found in most statistical handbooks.
I simply ask the following question: How many negative standard deviations are required to force the annual 3-year return down to the zero threshold? That number defines the probability of a year’s zero return reward. That probability merits attention.
To illustrate, say that the equity market had a 10 % annual 3-year average with a 15 % standard deviation. That means that only a negative two-thirds standard deviation event will drawdown my anticipated return to the zero level. That’s a meaningful target number. What is the probability of such a happening?
Using the standard normal distribution tables, a one-sided probability of exceeding that level is 25 % (accuracy beyond two places is sheer nonsense). Hence if the quoted values reflect my portfolio, I would anticipate a negative annual reward approximately 25 % of the time. That’s reasonably representative of the S&P 500 actual performance record.
Now that estimate is meaningful to an investor. He must be prepared to accept and/or reject and minimize that probability. Certainly one easy option is to design a portfolio with a much lower standard deviation expectation without sacrificing annual returns too much (free lunches are rare, but one exists in this arena). Portfolio diversification offers opportunities in this realm.
As another measure of risk, I would also be interested in the historical annual maximum drawdown that a portfolio, similar to my current design, experienced. This is one of many alternate measures of risk. You get to pick your own poison, your own risk measurement standards. A magic elixir doesn’t exist.
I hope this post contributed to the general and genial discussion. We are definitely not in a debate mode, at least not yet.
Best wishes and have a Happy Holiday season.
LA Times wine and dine critic's Holiday Books. HINT HINT Buy@ Amazon
http://www.latimes.com/food/la-fo-cookbooks-20131130,0,5379606,full.story
On a different note,today is Giving Tuesday.Please consider those not as fortunate as we may be.I will be giving to The Salvation Army,Partners in Health and Doctor's Without Borders.
As for the healthy debate, fortunately, that never ends.
Break, break.
Markets have stalled a bit lately. Awaiting next Fed move, direction of 10 year Treasury, and holiday spending, etc. And, surely, happy with the healthy advance this year...perhaps some folks taking money off table, or simply not jumping-in on the dip.
Trust all is well.
Thanks as always for your support.
Sorry for this late additional commentary. Since the originating postings highlighted the risk aspects of the investment process, my initial contribution also focused on that side of the investment equation.
When responding I solely emphasized the risk part of the conventional risk/reward tradeoff. I completely failed to document the detrimental impact that price volatility plays on net actual annual returns. We all understand that, because of the iron rules of mathematics, it takes a higher positive return percentage to recover from an unlucky negative return percentage.
A simple algebraic equation captures the difference between an average annual return and the actual annual return after correcting for the negative influence of a portfolio’s volatility. I apologize to those folks who hate equations, but, by necessity, here it is: actual annual return rate equals the average annual return rate minus one-half times the square of the annual standard deviation rate.
So portfolio or individual holding volatility (as measured by standard deviation) always detracts from the average annual returns. It is forever a drag on returns. For a portfolio that combines many investment components, the drag magnitude also depends on the correlation coefficients between all the individual elements.
Correlation coefficient is dynamic and changes over time and is dependent on market conditions. But it never gets to perfect correlation (absolute co-movement), and therefore always benefits a portfolio. When markets are challenged, these dynamic correlations do tend to gravitate towards the “perfect” level, thus attenuating its beneficial aspects. But they never completely disappear. That’s why diversification is such a powerful risk mitigation strategy; the tactic is indeed nearly a free lunch.
Let’s complete the picture by calculating the impact of volatility on the example that I provided in my earlier posting on the topic. With an annual return of 10 % and a portfolio standard deviation of 15 %, the portfolio’s actual annual return is 0.10 – (0.5 X 0.15 X 0.15) or 0.08875. Therefore, the actual annual return rate is degraded to 8.88 %. Over any period of time, the actual annual return rate is used when calculating accumulated wealth.
The investment goal should be to keep actual annual return rate at as high a level as your risk tolerance and practical return prospects permit.
High portfolio returns volatility operates negatively in both the returns dimension and in the risk dimension. High portfolio volatility erodes end wealth over time. High portfolio volatility increases the frequency likelihood of a zero annual return over a portfolio’s lifespan.
So work and monitoring is needed to minimize a portfolio’s overall standard deviation metric. From a practical perspective, cutting a portfolio’s standard deviation by half is an achievable target with only a small sacrifice in average annual return rate. Some of that sacrifice will be partially recovered by reducing a portfolio’s overall volatility.
Enough. Again, sorry that in my haste, I omitted a discussion of the influence that volatility has on portfolio performance. Thank you for your patience.
Merry Christmas.