Fund Selection for first timer

koru

Aegis wrote: »

Yes, but importantly, there was a much higher chance of a top quartile fund ending up in the top two quartiles than the bottom two quartiles.

xyy123 wrote: »

I never saw this mentioned in the study, actually. Simply because a fund is in the top quartile in year 1 does not mean it cannot be in the bottom quartile in year 2.

You need to look at the probabilities of the top quartile fund ending up in the top two quartiles in the subsequent period. The CRA2 report shows, for instance, that if you pick a UK equity income fund which has been top quartile over the last three years, there is a 33.5% chance it will be top quartile over the next three years and a 19.5% chance it will be in the next quartile down, so there is a 53% chance it will be in the top half. This means that by choosing the top quartile UK equity income fund there is a slightly higher than evens chance of beating the average. I am not sure I would call this a "much higher" chance. It means there is a 47% chance the fund will be worse than average. That seems pretty terrible odds to me.

The probabilities vary considerably between the fund sectors and depending on which period you choose to look over. You can pick from the CRA2 tables to find evidence on either side. For instance, picking the top quartile UK All Companies fund over the last three years would only give you a 48.4% chance of being in the top two quartiles over the coming three years. The best chances of picking a UK Equity & Bond Income fund which will be in the top two quartiles over the next three years would appear to come from picking a fund which is in the second quartile (counting from the bottom, therefore below average) for the previous three years.

Picking a UK Equity Income fund which is in the top quartile over the last seven years seems to give you a 78% chance that it will be in the top two quartiles over the next seven years. That seems like pretty good odds. However, the same seven-year strategy for UK Equity & Bond Income would only give you a 49.2% chance of being in the top two quartiles for the next seven years (in other words slightly worse than closing your eyes and sticking a pin in a list of all of the funds in the sector). I find it hard to believe there can be such a difference in the apparent persistence of outperformance between the two sectors and it makes we wonder if the analysis can be reliable (though I have no clue what their flaw might be).

Overall, looking at all the figures, the evidence seems to be that past superior performance gives a fractionally better than evens chance of future superior performance. On this basis, perhaps picking a wide range of active managers who seem to be the best will, overall, give a slightly better result than low-cost trackers in the same sectors, although nearly half of your active manager picks will probably turn out to underperform.

If it were possible to reliably spot which previous star managers were going to start underperforming, the odds could be improved considerably. However, for every Anthony Bolton, there's a Bill Miller. He outperformed the US stock market for 15 consecutive years, but since 2006 he has performed dreadfully. How many advisers recommended to their clients who were invested in his Legg Mason US Equity fund that they should disinvest in 2006 or even in 2007?

xyy123

koru wrote: »

You need to look at the probabilities of the top quartile fund ending up in the top two quartiles in the subsequent period. The CRA2 report shows, for instance, that if you pick a UK equity income fund which has been top quartile over the last three years, there is a 33.5% chance it will be top quartile over the next three years and a 19.5% chance it will be in the next quartile down, so there is a 53% chance it will be in the top half. This means that by choosing the top quartile UK equity income fund there is a slightly higher than evens chance of beating the average. I am not sure I would call this a "much higher" chance. It means there is a 47% chance the fund will be worse than average. That seems pretty terrible odds to me.

The probabilities vary considerably between the fund sectors and depending on which period you choose to look over. You can pick from the CRA2 tables to find evidence on either side. For instance, picking the top quartile UK All Companies fund over the last three years would only give you a 48.4% chance of being in the top two quartiles over the coming three years. The best chances of picking a UK Equity & Bond Income fund which will be in the top two quartiles over the next three years would appear to come from picking a fund which is in the second quartile (counting from the bottom, therefore below average) for the previous three years.

Picking a UK Equity Income fund which is in the top quartile over the last seven years seems to give you a 78% chance that it will be in the top two quartiles over the next seven years. That seems like pretty good odds. However, the same seven-year strategy for UK Equity & Bond Income would only give you a 49.2% chance of being in the top two quartiles for the next seven years (in other words slightly worse than closing your eyes and sticking a pin in a list of all of the funds in the sector). I find it hard to believe there can be such a difference in the apparent persistence of outperformance between the two sectors and it makes we wonder if the analysis can be reliable (though I have no clue what their flaw might be).

Overall, looking at all the figures, the evidence seems to be that past superior performance gives a fractionally better than evens chance of future superior performance. On this basis, perhaps picking a wide range of active managers who seem to be the best will, overall, give a slightly better result than low-cost trackers in the same sectors, although nearly half of your active manager picks will probably turn out to underperform.

If it were possible to reliably spot which previous star managers were going to start underperforming, the odds could be improved considerably. However, for every Anthony Bolton, there's a Bill Mason. Bill Mason outperformed the US stock market for 15 consecutive years, but since 2006 he has performed dreadfully. How many advisers recommended to their clients who were invested in Legg Mason US Equity that they should disinvest in 2006 or even in 2007?

Good analysis. But bear in mind also that the CRA2 study was comissioned by the IMA, and designed to show performance persistence in actively managed funds. The FSA said a large problem they had with the conclusions stemmed from the methodology, for instance using raw performance data rather than risk adjusted, which would have skewed the data in favour of performance persistence in top quartiles since over longer time periods, concentration of higher risk funds find themselves in the top quartile(s) (which I highlighted in my last post).

What this means is that a low risk (relative) fund may be performing extremely well for the risk it is taking on, but still find itself in the bottom quartile. Likewise a higher risk fund may perform terribly when taking into account the additional level of risk it is taking on and still be top quartile.

Thus, as you mention, seeing as there is a slightly better than evens probability of choosing performance persisting top funds concluded from the study and that the methodology used in the study is not fantastic, its not unreasonable to assume that the odds would be reduced if the analysis and methodology had been better.

koru

xyy123 wrote: »

But bear in mind also that the CRA2 study was comissioned by the IMA, and designed to show performance persistence in actively managed funds. The FSA said a large problem they had with the conclusions stemmed from the methodology, for instance using raw performance data rather than risk adjusted, which would have skewed the data in favour of performance persistence in top quartiles since over longer time periods, concentration of higher risk funds find themselves in the top quartile(s) (which I highlighted in my last post).

I wonder if that explains some of the apparent discrepancies in the figures? Perhaps the UK Equity Income sector is relatively heterogeneous as regards the level of risk taken by the fund manager, therefore the apparent high level of persistence of superior performance over seven years is, at least in part, a reflection of the higher risk funds tending to produce better returns? Whereas the UK Equity & Bond Income sector might tend to have similar levels of risk in all funds? I can't particularly see why the levels of risk diversity would be expected to be different, however. Indeed, if anything I would have guessed that the Equity & Bond sector might have more diversity, because some funds would be heavy on the bonds and some would be heavy on the equity, and over the long run one would expect higher returns from a fund which has a higher proportion of equity. (That is, over the very long run. Over the last 20 years, equities have underperformed bonds, although this is quite unusual and we are getting close to the all-time records for equity underperformance.)

What is, I think, pretty clear from looking at the various academic studies is that investing in closet trackers (active funds which closely mirror the makeup of the index but charge you 1.5% annual fee for the privilege) is almost certainly the worst option of all. Over the long run, this will be beaten by a portfolio of trackers or of truly active funds (which have a large portion of their assets invested in positions different from the market). The names Anthony Bolton and Neil Woodford are probably the ones most commonly mentioned by proponents of active management, and I think it is no coincidence that they both produced their excellent results by taking positions which were often contrary to the market.

Putting that another way, if you insist on going with actively managed funds, it is clear that you should be making sure you do not have a closet tracker. (It goes without saying that you should also not be investing in an actively managed fund which has been in the bottom two quartiles in terms of past performance. Such a fund has a better than evens chance of continuing to underperform.)

Arguably, this might point in the direction of looking at certain investment trusts, some of which are truly actively managed yet also have a TER similar to some of the lowest cost trackers. Perhaps the best of both worlds? There are of course other issues to consider with investment trusts, such as the extra volatility that comes from the possibility that they may trade at a discount or premium on the value of the underlying investments. Set against this price volatility, there are quite a few which have managed to sustain a consistently increasing dividend for more than 25 years, in some cases more than 40 years. That's the sort of record you cannot dismiss as being down to luck, although a consistently increasing dividend is not necessarily the same thing as consistently outperforming the total return from the market.

jamesd

koru, even getting a small chance of beating the average is a remarkably good result that can make a lot of money. Things that it didn't seem to consider that investors could use include:

1. Changing fund when the fund manager changes, to reduce the risk of underperformance from the new manager.
2. Changing to a different fund when the performance drops consistently for reasons that aren't attributable to economic conditions.
3. Changing to funds that are most well suited to anticipated economic conditions.

xyy123, you wrote "The FSA said a large problem they had" but you're citing a report commissioned by the FSA, not the FSA itself. There's a big difference between the official voice of the FSA and an external report it commissions to analyse some other reports. The analysis is interesting but it's still not "The FSA".

jamesd

giruzz wrote: »

10.4% ABERDEEN UT MGRS EMERGING MARKETS A ACC NAV
12.5% ALLIANZ GBL INV UK ALLZ PIMCO GILT YLD A INC
10.4% FIRST STATE INV ASIA PACIFIC LEADERS B NAV
12.5% INVESCO MGRS PERP HIGH INCOME INC NAV
12.5% M&G SECURITIES LTD INDEX LINKED BD A ACC NAV
10.4% NEPTUNE INV MGMT NEPTUNE EUROPEAN OPPS A ACC
10.4% NEPTUNE INV MGMT US OPPORTUNITIES A NAV
10.4% RENSBURG FD MGMT UK MID CAP GROWTH TST
10.4% THREADNEEDLE INV LATIN AMER INST ACC 2

Note sure that the 12.5% in gilts is a good idea when there's every reason to believe that quantitative easing has caused a significant price bubble in that market. I haven't checked whether M&G SECURITIES LTD INDEX LINKED BD also uses gilts but I'd be cautious about that one at the moment if it does use them a lot.

Asia-Pacific and Latin America have both had a very good year and there's reason to wonder if they are more set for a drop than additional large increases but those don't seem too unreasonable so long as you know this and are happy to accept the risk for the long term.

Some nice funds there assuming that you do want the high non-UK weighting that you have. I would want that, personally.

xyy123

jamesd wrote: »

xyy123, you wrote "The FSA said a large problem they had" but you're citing a report commissioned by the FSA, not the FSA itself. There's a big difference between the official voice of the FSA and an external report it commissions to analyse some other reports. The analysis is interesting but it's still not "The FSA".

True, my mistake. The important thing is its impartial and the FSA have deemed it important and accurate enough to endorse it, add their logo to it and publish it on their website.

koru

jamesd wrote: »

koru, even getting a small chance of beating the average is a remarkably good result that can make a lot of money. Things that it didn't seem to consider that investors could use include:

1. Changing fund when the fund manager changes, to reduce the risk of underperformance from the new manager.
2. Changing to a different fund when the performance drops consistently for reasons that aren't attributable to economic conditions.
3. Changing to funds that are most well suited to anticipated economic conditions.

Well, perhaps that is right, but it is a far cry from the much more bullish impression which is usually given by those who are in favour of active management. If you invest in a range of active funds, you are bound to end up with a few which turn out to provide long-term outperformance, such as Bolton or Woodford, but it is very unlikely that any adviser is going to give you a portfolio with a preponderance of such funds, which is the impression one is often given. (I am not pointing fingers at anyone in particular.)

Also, we need to take into account the fact that most investors give more weight to underperformance than over performance. This might not be rational, but it is human behaviour. Rationally, a 53% chance of over performance is a chance worth having, but how many investors would actually choose that if they understood that it means they have a 47% chance of underperforming?

In relation to your suggestions for improving performance of an active portfolio:
1 I completely agree with this. When Neil Woodford retires, it would make sense to sell his funds and wait to see if his successor has a similar magic touch.

2 If you can tell when performance has dropped consistently, then I agree. But at what point was it clear that Bill Miller had lost his touch? How do we know he will not regain his touch tomorrow? Neil Woodford has had lengthy periods of underperformance, but has turned out to be right in the long run, so far. In his next period of underperformance, how many years should we wait before we are sure he has actually lost his touch?

3 This supposes that it is possible to predict the future. There are a whole load of other academic studies which I understand show that there are very few people who can successfully and consistently do this. For instance, are we in a short-term rally, pumped up by quantitative easing, or will the rally continue? If you are sure you know the answer, then yes of course you should select funds accordingly, although you could also select trackers on the same basis. This time a year ago, I was feeling very pleased with myself for keeping a large chunk of money uninvested in the stock market, because I had felt from 2007 that the market was on its way down. Unfortunately, my apparent clairvoyance did not tell me that the market would suddenly shoot up from March this year and I have missed a lot of upside. This has taught me that I should not try to time the market.

Rollinghome

jamesd wrote: »

1. Changing fund when the fund manager changes, to reduce the risk of underperformance from the new manager.
2. Changing to a different fund when the performance drops consistently for reasons that aren't attributable to economic conditions.
3. Changing to funds that are most well suited to anticipated economic conditions.

James, there clearly is a possibility to select the best performing funds but doesn't that show just how difficult it is? And quite how difficult seems the hardest to quantify.

Several of your posts including this one http://forums.moneysavingexpert.com/showthread.html?t=1027133&highlight=cru from 18 months ago perhaps illustrate the problem.

For people who want fairly low overall ups and downs I tend to suggest investigating this mixture of investments at the moment:

30% BlackRock UK Absolute Alpha
20% Cru Investment Portfolio
20% Invesco Perpetual Monthly Income Plus
20% Invesco Perpetual Income
10% Neptune Global Equity

The Invesco Perpetual Income/High Income funds would perhaps still be seen as being two of the better active managed funds having been the top so long even though they now occupy the bottom three slots for the past 12 months. The jury may still be out on the high cost Blackrock Abs Ret fund. More serious would be a 20% holding in the now defunct Cru Investment Portfolio. With hindsight, which of those funds, if any, would still have been recommendations?

One of the dangers for the unwary seems to be the possiblity that by the time the performance stats persuade an investor to move its glory days are over. There's the real danger of constantly chasing performance only to arrive at the party too late.

koru

xyy123 wrote: »

Good analysis. But bear in mind also that the CRA2 study was comissioned by the IMA, and designed to show performance persistence in actively managed funds. The FSA said a large problem they had with the conclusions stemmed from the methodology, for instance using raw performance data rather than risk adjusted, which would have skewed the data in favour of performance persistence in top quartiles since over longer time periods, concentration of higher risk funds find themselves in the top quartile(s) (which I highlighted in my last post).

What this means is that a low risk (relative) fund may be performing extremely well for the risk it is taking on, but still find itself in the bottom quartile. Likewise a higher risk fund may perform terribly when taking into account the additional level of risk it is taking on and still be top quartile.

Thus, as you mention, seeing as there is a slightly better than evens probability of choosing performance persisting top funds concluded from the study and that the methodology used in the study is not fantastic, its not unreasonable to assume that the odds would be reduced if the analysis and methodology had been better.

I have just finished reading the full report, rather than just the quartiles tables. Although I did understand your point before, I understand it more profoundly now. The authors of the report only set out to prove persistence of performance within fund sectors, regardless of what was causing it. They themselves accept that their figures may not indicate skill on the part of the fund manager (what the economists call Alpha).

Arguably, the study is meaningless for the purpose of deciding whether picking top quartile active funds is a more rational investment strategy than investing in low-cost trackers. The study provides some evidence that investing in a portfolio of actively managed funds which have historically given top quartile performance can reasonably be expected to provide overall future performance slightly better than average for the sectors you have invested in, though the data fluctuates alarmingly wildly depending on the period of assessment and the particular fund sector. However, even this marginal tilting of the odds might be purely attributable to the likelihood that this strategy will be biasing you towards selecting funds which take a higher risk than the market average. If so, then if you are willing to accept higher risk for higher returns, there are plenty of ways to increase your potential long-term return using trackers (at the expense of high risk), for instance by selecting a smaller companies tracker or emerging markets tracker.

It is a shame that CRA did not carry on their work and look at risk adjusted returns, or that their client, the IMA, did not ask them to do so.

koru

By the way, deepest apologies to the original poster for perpetuating the discussion of points that are not especially relevant to the original query. However, this thread has turned into an exceptionally useful discussion about trackers versus actives and has certainly helped refine my understanding of the real points on both sides. Too often, proponents of either side simply assert that it stands to reason that trackers/actives are a better bet, but this thread has got into a lot of nitty-gritty evidence. Unfortunately, the evidence is not entirely conclusive, but it is good to understand it better.

Incidentally, it has been contended that most of the US studies are irrelevant in the UK because trackers have a tax advantage in the US which will tend to help them seem to perform better than active funds. However, most of the US studies I have seen seem to compare actively managed funds with the index and with the average performance for the sector, rather than directly comparing with trackers. In the case of these US studies, tax should not distort the findings. For instance, one study finds that the number of US active fund managers who outperform the market is pretty much what you would expect as a result of pure luck.

I will not pretend to understand the detailed approach taken in the study that reaches this conclusion, but as I understand it the authors used the following principle. If you have 1000 children who claim to be clairvoyant and you ask them each to toss a coin (an unbiased one) 8 times and challenge them to predict heads or tails, statistics would suggest that roughly 4 of them are going to end up with 8 correct guesses. That does not mean those particular children have any real powers to predict the toss; they were just lucky. If 20 of them guess right 8 times, then this suggests some of them are indeed clairvoyant (though about four of them are just lucky). As I understand it, the authors of this study used statistics to predict how many of the population of US fund managers would be expected to outperform due to pure luck and this pretty much accounts for actual recent performance of fund managers.

The paper is available by clicking on download here:
papers.ssrn.com/sol3/papers.cfm?abstract_id=869748&rec=1&srcabs=967553

The implication is that there is no reason to think the luck of the outperformers will hold. If the four children who predicted 8 times in a row toss again, they each have only a 50% chance of managing another correct guess. Maybe Bill Miller was just one of those children and his luck ran out; maybe Neil Woodford is one.