Effect of MER on Performance, Reversion to the Mean

M = MER
r = published return
R = real performance (before MER expenses) = (r+M)/(1-M)
Example: M = .02, r = .2, R = .224 (or 22.4%)
or, Q-a-D. approximation:
R = r+M = 22%

... just had milk and cookies with U-know-Who ...

Or to put it another way: Assume an original value of $100 which would grow to $120 after 20% gain.

Real return = (Appreciated net value/(100%-MER))*100

So, R = (120/(100%-2%))*100 = 122.45

I was awake about the same time as Gummy trying to figure this out but had to wait until now to post.

Merry Christmas to all.

Billy, a simpler and less mathematical way of thinking about this topic is:

20% return posted to investor + 2% MER = 22% real performance prior to the deduction of expenses.

This simplicity is for those of us who are mathematically challenged. :-)

Katie

You sure took the fun out of that exercise!!

For another perspective on the question, let's look at the MER as a percentage of a fund's total return. Historically the average equity fund has returned an average of about 10% per year. So an MER of 2% represents about 20% of your total investment return (i.e. performance) over the long haul. BTW, the average Canadian equity fund charges an MER of more like 2¼% to 2½%.

Something to think about the next time someone tries to convince you that MERs aren't important.

The fees charged by a fund manager and related expenses are the cost of doing business. By that I mean that most people do not have the education or do not posess the time to manage their own portfolios. Therefore, these people pay the MER etc. and as long as the returns justify the fees, MER's are of little importance.

HBH:

Now there is a common fallacy. If you take a look at long term performance of a group of similar funds you will notice a regression to the mean, in performance. You will also notice that the hier performers tend to correllate with the lower MERs. This is especially true for large cap funds.

What you pay to buy and stay in a fund (loads and MERs) is of crucial importance to long term performance.

I should also point out that low-MER index funds should also be a core holding (the growth-large cap component of a portfolio). Over the long term, large cap fund management (of a particular style) doesn't seem to add any value.

To: Brian Gomke:

I've read a great deal about "regression to the mean" but I have yet to find a rigorous analysis of the (alleged) phenomenon. Bogles' work (Bylo are you there) draws very hgeavily on the concept, but I can find no reference to it in textbooks on statisitical theory. My own view is that it is pseudo-scientific nonsense, imbued with legitimacy by people who don't want to take the time to examine the intellectual underpinnings of an argument. It is absolutely true that the long-term average returns of the market are remarkably consistent (at about 10%) but people don't buy the market (unless they buy an index fund). They buy the (good or bad) talents of a money manager, and while some (most?) are bad, a few are very good. Buffet, Templeton Lynch come to mind immediately. (Or do you subscribe to the view that these guys are really just lucky SOBs who've fooled the rest of us.)

Bottom line: it is possible to beat the market over the long term (although it is very difficult) and because it is, regression the mean is (sorry, but I have to do it.....) meaningless.

I wonder just how the big dudes collecting the riches for their efforts would feel if, the fees they where paid were based on returns. I just bet there would not be so many hot-shots out there.

I also take issue with the comments anyways that it is one person's direction. I know they love to take the credit when all is well but when the fund suffers, all of a sudden it was the research team that messed up. Look at the head Honcho (Startz, an extremely talented gent) from Global Strategy. All the Ballyhoo didn't make a damn nickle so they reduced his exposure so he could work wonders in 'his field'. ??

thbox--- loved your response to Gomke. One should pay attention to the MER of a fund but it should not be the overriding issue. There are several great funds with low MER's, PH&N come to mind, but there are also funds with high MER's which have long term performance which justifies the fee. Templeton Growth I believe has an MER of 2.00 and has one of the best long term performance records of any fund in its class. On the other end are fund with extremely high MER's and crappy returns like Apex Equity Growth with an MER of 3.00 and a 3yr avg return of 7.9%. I'm sure there are many more examples. The most important issue I believe is to get people to invest instead of giving their money to the banks!!!

thbox: I'll see if I can dig up some data to validate (or invalidate) the application of regression to the mean to mutual funds.

Gummy:

I would be most interested.

I'm also looking for more info. Neither Bogle's book nor Malkeil (A Random Walk Down Wall Street) reference the term "regression to the mean" -- at least not in the index.

Bylo:

Bogle on Mutual Funds refers to "regression to the mean" at the following pages: 20, 87, 92, 174, 177. There may be others that I missed.

...but not in the index! :-(

about Regression to the Mean:
I think the idea is that funds in the top quartile are there because they had some unusually good years (compared to the average fund). In future, the returns for these funds will undoubtedly fall (toward the mean).
Funds in the bottom quartile had some unusually bad years. In future, the returns for these funds will undoubtedly rise (toward the mean).
I'm not convinced this argument applies to mutual funds (as it might to scores on midterm tests compared to the final exam) ... but do you think I can find year-by-year returns for a variety of funds?!@#$ Aaargh!

... but I'm still lookin'

Gummy, if you can get your hands on the book "Understanding Mutual Funds" by Steve Kelman there are yearly returns going back to 1978. This book is part of the Globe and Mail Personal Finance Library.

(I was one of the lucky people who got a complementary copy from Mr. Kelman because I answered another thread awhile back on mutual fund books. This book is definitely not a piece of fluff. It has lots of information, charts, etc for the serious mutual fund investor.)

Gummy:

The Globe publishes its fifteen year annual return data in early February of each year. I have data that goes back to 1974, but its hard to manage because the fund names etc change so much. I have a copy of Kelman's book (the 1993 edition) but his data isn't calendar year which makes comparisons impossible. (I'm not much on Kelman's book anyway)

Your essential point - that dramatic outperformance can't be sustained is absolutely valid. The more fundamental issue - whether outperformance of any kind is sustainable (which is denied by those who believe in "regression to the mean" is much less amenable to simple answer. I believe it is - because no one has yet offered me an explanation for why Buffett, Lynch (or in the mutual fund world, Templeton) can do so well. (Dumb luck is NOT an adequate answer.)

As a observation only, the thread has moved somewhat from outperformance and MER, but hey, I'm up for stream of consciousness debate

Thanks, all, for the leads. I think I may be able to come up with something. If (when?) I do I'll get off this MER thread and start a Regression thread with beeyutiful pictures, mebbe some math 'n such.

... just finished sippin' coffee and watchin' the PalTrak demo download. Mamma mia! There's enuff info there to keep me out of my wife's hair till Y2K!

and I hafta look at Kelman's book ...

... jest gotta make one more comment to thbox (then off to bed):

If Regression to the Mean really does apply to mutual funds, mebbe ah'll start me one, pick stocks via the dart board technique and let this here statistical phenomenon lift me happily to the mean.

gummy,

Isn't that what an index fund does? Of course, the dartboard adds fun to the fund.

thbox,

I tend to agree with you that there has to be more than blind luck at work for a selected few like Buffet, Lynch, and Templeton (but what about Frankie Mersch?)

Bylo:

What about Frankie Mersch. There is a BIG difference between skill and "madness of crowds". Mersch IMHO is a classic example of the power of marketing. Like others, he is protrayed as a "star" (ditto Hirsch, Zechner, Baird, .....) on the basis of relatively short-term performance. Notice that Mersch's antecedents are shrouded in mystery. Sure he was at Morgan Trust before he went to Altamira, but what exactly did he do there. What was his "record"? Kiki Delaney is another. She was at Gluskin Sheff prior to setting up on her own, but what was her performance record at Gluskin Sheff (or did she have one)?

I know that statistically the time period required to distinguish luch from skill is very long (30 to 70 years depending on who's studying). Nevertheless, I believe it's still possible to get a sense of the difference on the basis of shorter times (say 15 years). That's why I like Krembil so much - you can actually see his performance since the mid-70s at the old BT Taurus Fund (now sadly merged and gone).

One of the really intractable difficulties in this sort of analysis is how you build in the "succession" question. Templeton's fund have done very well over the past 40+ years, but the recent performance is more properly attributable to Holowesko. Does he get the benefit of the Templeton track record? I think so, but my reasons are very ad hoc.....

Sorry for the ramble. Thanks for the hot link.

thbox,

I purposely included Frankie in jest. Up until a couple of years ago he seemed to possess the same "brilliance" as the other all-stars in the list. Now the $64,000 question (more/less depending on the size of one's position in Altamira Equity): is Frankie merely in a temporary slump? For the answer, or at least lots of speculation, check out any of the many recent threads dedicated to that topic.

As for Templeton, Sir John deserves special mention not only for being an all-star fund manager in his own right, but also -- and even more importantly -- for establishing the system/discipline that has enabled subsequent managers like Hansberger, Holowesko, et al to perpetuate his outstanding performance record.

Now that we've all been infected by the Asian flu it's especially worth listening to the audio tape of Templeton Growth's annual meeting in July. (I believe you can get a free copy by calling Templeton.) Those guys sure seem to know what they're doing. Growth got through the flu largely unscathed and Emerging Markets is still down maybe 10%. Compare that to some of their peers.

And yes, Krembil has established a similar management system/discipline at Trimark.

That's why I continue to like (and own) funds from both families -- yes, even despite the tax sting from the recent Trimark distributions! ;-)

Bylo:

I knew it was jest. Did so, did so did so.

There was a recent article about Mersch in either ROB Mag or Saturday Night, in which Mersch whined about being written off by the investing literati. IMHO, Mersch is a very smart guy whose investment approach is unsuited to the size of his fund. When you're managing $200M, you can buy (and trade the bejeezus out of) the Vengold's and Cartaways of the world with some effect. When you're running $2B, you need to be right about a whole lot more Vengolds and Cartaways to have the same effect. So Mersch pooped.

In early 1995 or 1996, he recognized the difficulty, and started to modify his approach to reflect the new reality of the fund. (This involved (inter alia) more bank stocks for liquidity and stability.

So back to the issue.....I think Mersch may be able to bounce back, but it won't be on the basis of his old approach. And because he has no track record with the new approach, I wouldn't buy it - regardless of how smart he is.

On the issue of discpline and consistency, I agree with you completely. How though do you rigorously evaluate a fund company's discipline, particularly through a change in personnel? We can do it after the fact, but before the fact is quite a different issue.

Just to "spice up" this debate, I suggest that "regression to the mean" does not exist at all. This statement implies something backwards to that which is actually occuring.

What actually happens is that "the mean" regresses to the compendium universe being measured.

Kinda reminds me of the "efficient markets theory" which Buffett blows apart in his 1988 Chairmans letter. Flawless logic, proven by example.

thbox your statement of

>Your essential point - that dramatic outperformance can't be sustained is absolutely valid. The more fundamental issue - whether outperformance of any kind is sustainable (which is denied by those who believe in "regression to the mean" is much less amenable to simple answer. I believe it is - because no one has yet offered me an explanation for why Buffett, Lynch (or in the mutual fund world, Templeton) can do so well. (Dumb luck is NOT an adequate answer.)<

is suggested by those statistically "in the know" that their record simply reflects the normal dispersion pattern inherent in any statistical measurement. jd showed it on a thread a while back.

I personally think it's hooey that no one can outperform the markets over the long term.

The reason that Buffett is able to continue to outperform hinges, IMHO, on the following factors:

1] Buffett values all companies using an "absolute" method (DCF) as opposed to the more conventional "relative" valuations (i.e P/e's, Cash Flow multiples etc).

2] Requires a "margin of safey" on all purchases. Won't buy unless the security is priced 'significantly' below the inherent value of the security.

3] Buys companies with a high return on equity.

4] Focusses on companies with *consistent* earnings growth, making proper valuation easier.

5] Focusses on companies which have brand power (a "franchise") and/OR which are the low/lowest cost providers in their industry (particularly important if they are in "commodity" oriented businesses).

6] Use of "free" leverage through the consistently profitable insurance operations (i.e. combined ratio almost always below 100%, a rarity in that business).

7] His willingness to buy enough securities to make a meaningful difference to his portfolio (which Lynch calls "backing up the truck"), when all the other factors have "lined up" (In other words, he's identified an undervalued security he wants to buy).

Lynchs' success was based on a different formula, which can be partially explained by his propensity to buy small cap stocks (which traditionally outperform large caps), and also by his willingness to "back up the truck" when a huge opportunity was identified.

Hope this helps.

Cheers!

Jay

Oh, two other things re: Buffett outperformance.

8] Little capital gains taxes, due to length of time holds securities (he estimates that has added up to 2% p.a. to his return rate) and

9] His management costs (equal to an MER in this instance) are razor thin at only 2 basis points, compared to a conventional 200 to 250 basis points for most mutual funds.

All these factors add up to outperformance.

Cheers!

Jay

...because no one has yet offered me an explanation for why Buffett, Lynch (or in the mutual fund world, Templeton) can do so well. (Dumb luck is NOT an adequate answer.)

I agree that dumb luck is an inadequate answer, however, random chance is perfectly adequate. Occam's razor suggests that the simplest answer is usually the right answer, so I contend that random chance is responsible for Lynch et al. He is nothing more than an outlier - a very smart outlier, but an outlier nonetheless.

Here's some statistical observations that support, not prove, my contention - nothing can ever be proven with the aid of statistics. Random chance suggests that the odds of flipping a coin 5 times and coming up heads every time is 0.03125 or 3 percent.

If the performance of a CDN equity mf mutual were a random phenomenon then around 3 percent of all CDN equity funds with at least a 5yr history should have returns in the 2nd quartile or better over the past 5 years. So how many CDN equity funds have actually beaten the median over the past 5 yrs? (I am using dec1992 to dec1996 statistics compiled by Paltrak) There are on record 127 CDN equity funds that have at least a 5yr quartile ranking so if fund performance were governed by random chance we should anticipate 0.03 X 127 = 4 funds meeting such criteria. According to Paltrak's filtering criteria, 1 fund meets the above criteria - Empire Group Equity - which is roughly what we'd expect if mf performance were a random event.

Applying the same approach to Int'l Equity funds, random chance would suggest that around 1 or so funds would perform above-median for 5 straight years. Paltrak spits out 2 funds meeting the criteria.

So while this doesn't prove that fund managers don't make a difference, that fund performance is simply a random event, it does point seem to point in that direction.

To jd:

I agree entirely that outperformance is conceivably a random outcome (I called it dumb luck merely for esthetic reasons). I think though that 5 years is much too short a period over which to measure the effects of "skill" vs. "chance" The study drawn to my attention by Bylo does a very good job of detailing the distinction in various categories of US funds, and Bernstein (the author) uses 11 years as the basis. As a said before, I've read that only with very long track records (30 to 70 years, depending on the study) is is possible to distinguish skill from chance with any meaningful degree of confidence. My own view remains that track records of 15+ years are a good start.

There are two difficulties in your analysis. The first is that your universe of funds does not distinguish funds which have achieved their records under a single manager, and those that have done so with a variety of managers. In effect, you're evaluating the fund's skill, rather than the manager's but it is of course the manager who is at issue. I don't care how great the track record of a fund is if it was achieved by someone who has departed. (AGF's Cdn Equity Fund is the one that comes first to mind, but there are multitudes of them.

The second is that you (appear to) see performance as a series of discrete one year contests. IMHO, it doesn't matter if the manager underperforms the index once (or twice) in a long-term record, because the long-term compound return can nevertheless exceed the average. If you want to use 5 year data, you should at least look at rolling 5 year compound returns. To get five years of data, that would presuppose a ten year track record.

Can someone please explain the difference to me what the difference is between a MANAGEMENT FEE and a MANAGEMENT EXPENSE RATIO.......I notice that the management fee is usually lower than the MER...

Thanks!

The management fee is the fee charged by the manager to run the fund. The management expense ratio (MER) is the total of the management fee, and various other costs associated with running the fund, including custodian fees, legal fees, accounting and auditing fees, etc. Sometimes the propsectus of the fund provides that these other expenses are to be paid by the manager out of the management fee, but that doesn't happen very often. More frequently, marketing and promotional fees are paid for out of the management fee.

Where the fund's portfolio is administered by an external agency (like Specturm United (the manager) retaining C.A. Delaney & Assoc. to manage the portfolios of a couple of Cdn equity funds), the portfolio adviser is paid out of the management fee. Depending on the kind of fund it is, that sub-adviser fee can range anywhere from 20bp to 1+ %. The balance of the management fee pays for client adminstration, etc.

I've left out brokerage fees. Aside from the management fee, these are the most significant expense of the fund, but for accounting reasons these are usually included in the "cost of securities" in the financial statements.

The other big cost for funds is distribution. On a rear load fund, the broker is paid (up to) 5%. The fund company doesn't get that money until the investor cashes out (if at all) but they have to pay the broker up front. They do it through limited partnerships which are entitled to favorable tax treatment (at least till the end of '98). The partnerhsips are paid a portion of the fund's management fee, and that revenue together with the revenue stream from redeemed units is used to pay the broker. It works only because of the tax incentives, and when they end, the fund companies will be scrambling to find another way to fund DSC funds.

Although some portion of the management fee is used to pay the DSC, the MER doesn't include the sales charge you pay. It isn't that linear. However you've probably noticed that DSC funds typically have a higher management fee than front-load funds. That's because the management fee on the DSC funds has to include some amount that is paid to the limited partnerships to fund the DSC.

Hope this helps, and sorry if it doesn't seem straightforward.

WOW! Thanks very much thbox!

I've read that only with very long track records (30 to 70 years, depending on the study) is is possible to distinguish skill from chance with any meaningful degree of confidence.

What sources are those? If it takes 30 to 70 years to distinguish a difference between skill and chance the differences must be very minute indeed.

My own view remains that track records of 15+ years are a good start.

Why? For sound empirical reasons or because it holds out hope of vindication of your preferred belief?

There are 67 Cdn Equity funds with a >= 10yr track record. If we are assuming that fund performance is a random event we would anticipate the odds of a fund with an above median performance to be 0.098 percent. That would imply we should expect 67 x 0.00098 = 0.065 funds. No fund meets it, as random chance would predict. There is no point going on with 15 yr history since we know a priori that no fund taken from an even smaller sample will meet the even more rigorous criteria. Sure, it would be nicer to have more data to provide more robust summary statistics but that isn't possible. In time it will come but all we can do is analyse the existing data. And the current data suggests that if fund performance isn't random it certainly looks like it is.

There are two difficulties in your analysis. The first is that your universe of funds does not distinguish funds which have achieved their records under a single manager, and those that have done so with a variety of managers. In effect, you're evaluating the fund's skill, rather than the manager's but it is of course the manager who is at issue. I don't care how great the track record of a fund is if it was achieved by someone who has departed. (AGF's Cdn Equity Fund is the one that comes first to mind, but there are multitudes of them.

I performed a simple test. I figured out what the probability was for a perfectly random event like sequential coin-tossing. I then compared their associated probabilities with the actual probability of a Cdn Equity fund performing above median sequentially for 5 and 10 yrs taken from the population of 67 and 56 funds, respectively.

And what fund manager manages a fund for the requisite 30 to 70 yr period that you feel is necessary to distinguish between skill and chance?

The second is that you (appear to) see performance as a series of discrete one year contests. IMHO, it doesn't matter if the manager underperforms the index once (or twice) in a long-term record, because the long-term compound return can nevertheless exceed the average. If you want to use 5 year data, you should at least look at rolling 5 year compound returns. To get five years of data, that would presuppose a ten year track record.

I didn't draw comparisons with the tse index. I simply observe the annual performance of the population of Cdn Equity funds out there and compare it with a simple random event like a coin toss. The behavioural similarities are difficult to dismiss.

Looking at compounded returns is less informative than annual statistics. Looking at annual data provides a more informationally complete picture of overall characteristics. One can always determine the compounded returns if the annual returns data is available, but not the converse.

jd & thbox,

Wow, were you guys on the debating team in high school? - some pretty good to and fro there!

Here's another little nugget to ponder:

Just because there are only a very limited number money managers are actually able to outperform their benchmarks over the long term (net of expenses) doesn't mean that the people who accomplish this feat have done so by 'random chance' because probabilty theory states that, given such a large representative sample, 'somebody will likely beat the market'. In other words, money management isn't necessarily akin to the random chance of say, winning a lottery or flipping a coin such that heads comes up five times in succession.

Take the case of the National Basketball Association (NBA) in the U.S.. I believe they have approx. 30 teams and that each team has about 10 players, for a total of about 300. Now, consider that segment of the male population in the U.S. who are:

1) skilled enough in basketball to start on their senior high school team

2) have a desire to play in the NBA

I think it would be fair to say that this 'pool' likely represents 100s of thousands of young U.S. atheletes, given the popularity of BBall State-side.

Now, think of what it takes for these kids to make it into that 'elite' 300 (people who can see where I'm going with this, raise your hand...):

- must be talented enough at the high school level to attract the attention of college scouts, and be seen as valuable enough to be offered a scholarship with an NCAA team.

- make the college team

- start for the college team

- star for the college team to attract the interest of pro scouts

- be such a phenominal pro prospect that, out of the group of players representing the 100s and 100s of college teams that play basketball in the U.S. each year, be one of the 60 (that's right, SIXTY) players taken in the NBA draft the year he becomes eligible (2 round draft in the NBA).

- make the team that drafted him (or another NBA team)

- 'survive' in the NBA (eg. stay injury-free, fight of new recruits brought in to challenge for your position each year, etc.)

Even if a person make it though all this, he STILL might only be an average NBA player (eg. a reserve, a borderline starter, etc.).

Obviously, no one could seriously claim that stars in the NBA got there because 'somebody has to play pro basketball' or 'ranom chance was on their side': they are there because they are the best in the world at what they do and have a tremendous amount of skill in this area. I'd argue that the same goes for mutual fund managers who outperform over the long term.

To further the analogy, I like to think of us, do-it-yourselfers, or financial planners/advisors as assuming the role of the pro basketball scout: our job is not to look to identify the funds that have performed well in the past (if it were that easy, anyone could look for the 'big numbers' in the return columns, buy those funds, and go to sleep - it's just like a scout reporting to a GM "y'know, that Michael Jordan is a great basketball player"). Rather, we use the experiences we have gained from the past and apply what we know to identify those funds/other investment instruments which we are confident will help us achieve our investment goals. similar to scouts picking their favourite prospects for their respective NBA tams, we too must identify those fund manager 'prospects' whose philosophy an style jive with our beliefs. Or, if you like, just scout scours the league in an attempt to find exisiting players that might help his club via a trade or free agent signing (trying to decipher things such as whether the past productivity of the player will likely continue in the future, if the player was likely a one-year wonder, etc.), we too must look under the hood of a fund managers (fill in your own joke here) or individual fund to determine whether or not we can feel justified in being confident that (s)he'll produce in the future as (s)he did in the past. Bottom Line: in both instances, I'd argue that it's the skill of the 'players' involved, not probability theory, that determines who the 'blue chippers' are.

jd:

I'll reply more fully later. Charles Ellis's "Investment Policy: How to Win the Loser's Game" quotes one Barr Rosenberg on the need for 70 years worth of data (at p.63). Bogle on Mutual Funds quotes one R.A. Brealey: "you probably need at least 25 years of fund performance to distinguish at the 95% confidence level whether a manager has above-average competence." Bogle then cites another unnamed commentator to support the position that 25 years of data will work only if the benchmark is perfect, and that a less than perfect benchmark might increase the observation time to 80 years.

I don't agree with much of Bogle's view, but I don't ascribe intellectual dishonesty to him. I assume that the "unnamed" commentator exists. So in fact I got the short end of the horizon wrong - its 25 years.

More to follow. Thanks.

I forgot one thing. The reference in Bogle is at p.177.

jd:

A final reference. Jeremy Siegel, "Stocks for the Long Run" (at p.289) contends that after 30 years, the probability of distinguishing whether an additional 1% return is the product of luck or skill is only 67%. To distinguish luck from skill over 30 years to a probability of 95%, the manager would have to outperform by 4% per year.

That's all for now, but I'll resume tomorrow.

regards

thbox, send me an e-mail and I'll send you a WORD page showing how to put a hot link in a posting.

Cheers!

Jay

Jay and thbox,

Colt,

Believe whatever you want. I just report the numbers and point out the similarities between a fair game of chance and the number of stock-picking winning mf managers. It is difficult to discern the difference.

I don't think the comparison between mf managers and professional athletes is a particularly apt one. Professional athletes control the game, they are the game. MF managers are simply observers of the game and lay odds on its outcome - like bookies. They have no control over the game's outcome. And the outcome is totally unpredictable with all the unknown, exogenous random shocks that influence markets throughout the year.

Here's another interesting summary stat I ran across on Paltrak the other night. In one, three, five, ten and fifteen yr periods what proportion of CDN mf's were beaten by the tse? For all time periods the answer was somewhere in the area between 65 and 75 percent. If the performance characteristics were randomly distributed we would anticipate the proportion to hover around 50 percent. The only answer I can come up with to explain this distinct skewness is the MER of the funds.

jd,

The main point I was trying to make in that (admittedly lengthy) monologue was that just because only a select few fund managers are able to outperform over the long term doesn't necessarily mean their outperfromance can be attributed to random chance.

And the quest for truth rages on.....

"MF managers are simply observers of the game ... They have no control over the game's outcome."

Take Frank Mersch and Veronika Hirsch, puhleese.

jd, what happens to those stats if the average MER of 2.15% (according to the latest 10 year figures I have available) is added back into the MF. Does the average then beat or lose to the TSE (300?)

Jay

I don't believe all of these stats on randomness and averages mean much. When you think about it, the overwhelming majority of people have more than the average (mean) number of legs. Which is to say that sometimes these numbers are meaningless. ;-)

jd:

Sorry, there are no hotlinks in this - maybe next time, and thanks again for your help.

Let's start with your final set of propositions - that annual data provides an informationally more complete picture of overall characteristics. First, I fail to see why 12 month returns based on calendar years is somehow more significant and "informationally complete" than 12 month returns based on some other fiscal year. I recognize the argument that if a manager doesn't beat the average consistently over 10 calendar years, there is no point in looking at (say) Nov - Nov returns, but if pedantry is your preferred style, you need to insulate your argument a little better

While I agree that compound return obscures a good deal of information, it doesn't follow at all that it is always less useful than compound returns. You're right that you can calculate CAR from annual returns, and not vice versa, but so what. That fact doesn't bear on the argument for, or against, the utility of compound returns in the context of evaluating long-term performance. The utility of CAR is precisely that it allows easy comparisons of returns over periods greater than one year. Since my basic assertion was that it is a manager's long-term record that is relevant, rather than his/her performance over discrete annual periods, I don't see how looking at annual returns rather than CAR might advance the process. You can slice and dice the data differently if you want, but I'm impressed by a manager who manages to beat the index (or the average - take your pick) consistently over 10 year periods. The criterion of beating the average consistently over calendar years seems unnecessarily demanding. Its sort of like acknowledging that even the best managers can make a mistake every now and then without entirely denying their skill.

You want to make a great deal of the analogy between coin-flipping and portfolio management performance. This sort of activity was popularized by Burton Malkiel (of Random Walk fame), but it is worth more than mere mention that Malkiel has largely renounced his efficient market hypothesis (EMH). There are two intellectual threads here.

There is a world of difference between flipping a coin and portfolio management. (I suspect Vito Maida is wishing just about now that the difference wasn't so grand.) In coin flipping, there are two outcomes: head or tails; black or white. In the real world though the Great Designer (I wonder if he's related to Versace) likes to paint in shades of grey. As Brimelow points out in his analysis of investment letters (a la Hulbert),

....EMH proponents have a suspicious tendency to vagueness when it comes to specifying to what exactly their coin flip is analogous. Is it the odds of a successful stock pick…or the chance that a whole portfolio will appreciate? If the odds of the former are 0.5, the odds of the latter, assuming equal allocation between the stocks must be 0.5 RAISED TO THE POWER OF HOWEVER MANY STOCKS ARE IN THE PORTFOLIO. This makes an appreciating portfolio much more of a feat. Similarly it makes a vast difference to the probability calculation whether the coin flip represents price action over a day or various combinations of days, such as a year or five years….

One of the most compelling arguments against EMH is that the past movements of the market as a whole can provide no more evidence of the market's future movements than the past movements of individual securities can of their future movements. Yet the evidence OVERWHELMINGLY indicates that the LONG TERM movements of the market exhibit remarkable stability. Over the past two hundred years, the average return on equities has been 10%, or to put it another way, the long-term risk premium on equities has been remarkably stable at (about) 7-8%. How come? Why is it that the "market" isn't as smart as its constituent elements?

You are correct - 30 years of data is difficult to come by. The most commonly referred-to case is that of the Value Line Investment Survey, which has operated since 1965. The consistency of this survey's ability to identify investments which outperform has been remarked on numerous times, and even Nobel Laureate Fischer Black (of Timmons fame) - who is an avowed disciple of EMH - concedes his amazement. Does one instance make a case? Of course not? But since the argument of EMH is that the market cannot be beaten (on a risk adjusted basis) over the long term, it is clear to me that EMH is wrong, even though the methodology for identifying the prospective winners is rudimentary or deficient, or both.

My view, and it is only my view, is that the best way to identify propsective outperformers is to look at a MANAGER'S record, for its discipline and consistency rather than the simple scope of its past outperformance. Consistency is hard enough to quantify, and discipline is even harder, but the methodological difficulties do not preclude the validity of the hypothesis. As I indicated at the outset, I continue to believe that a track record of 15+ years is a good start. As a betting man (you must be, if you think coin-flipping reveals useful data), you have to acknowledge that the probability of outperforming over 15+ years by luck alone is lower than the probability of outperforming over 3 years. Bill Berger has cautioned: "Don't confuse brains with a bull market" and in the light of the past five years, the ability of a manager to weather the bears of '87 and '90 (never mind 94 - it wasn't big enough) IS important, despite the disparagemernt of the coin flippers of the world….

On a lighter note, I was particularly fascinated by the statistical argument that since the overwhelming majority of people have more than the median number of legs, statistics must be a pseudo-science. I suspect he used both of his to come up with that view….

regards for now