Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Hi Irwin,
That looks like an interesting book. Thanks for the link!
Hi Conrad,
So I guess there'll be no "Conrad, President for Life" anytime soon?
Hi Patrick,
It's nice to see more and more sites dedicated to looking at risk rather than just chasing returns.
Thanks for the link.
Hi Conrad,
Arnie might become Governor of California, but I don't believe he's eligible to be President (since he's not a natural born US citizen).
But it will be interesting to see how far he goes.
Hi Irwin,
I hadn't seen that site before. It looks like they have two major levels of service. I'm not sure how the personal level works, but the institutional level appears to use some sort of MPT algorithm.
Hi J,
"What I mean by diversity is a personally-evaluated risk-tolerance judgment of a proper division between equities and fixed income with the equities divided between Large cap growth, Lg. cap value, Small cap growth, Sm. cap value and a significant exposure to international funds or ETFs, at least."
Yup, that's a pretty good definition of diversity. Diversifying that way will lead you to hold components with low correlations and thus I'd agree with your statement that, "asset diversity is crucial for optimal outcomes."
However even if we diversify in a less than perfect manner, while we might not see optimal outcomes, I think we would nonetheless see benefits with MPT.
"My idea is to AIM each class and let AIM do the rebalancing work, since AIM automatically sells advancers and buys decliners, and forget the yearly rebalancing."
I'm still not sure I see what you're getting at, however I'll make a guess. If my guess is incorrect, then if you'd provide an example, I'll provide my thoughts.
So without further ado, here's my thoughts on what I think you're asking...
If your idea is to diversify, as above, and then let AIM manage each ETF (or whatever product you used to implement your portfolio) then I don't believe you could get away with not rebalancing at some point.
Over the long term your different asset classes will perform very differently. Some might generate an abundance of cash as they consistently climbed higher while others might simply have a reasonable equity/cash ratio.
This means that the high-fliers would, at some point, be under-represented in your portfolio. Therefore you'd have to rebalance if you wanted to get back to reasonable ratios.
I think AIM is good at managing the cash/equity ratio for a single equity (or a group of equities that behave the same), however AIM doesn't know how to manage risk when it comes to multiple equities that behave differently. It treats each equity as an isolated, independent island unto itself with no regard for the others.
This, in my opinion, is not a good way to manage an overall portfolio.
"I`m taking into account... an expected amateurish reluctance to sell winning assets and reinvest the proceeds to the lagging sectors as MPT demands that you do."
It seems to me that you already do something similar to this with AIM, so I don't think it's too big a jump to do it with MPT.
The difference, of course, is that with AIM your cash is put into distinct compartments that aren't usually transferred between equities. However in reality, these compartments are just logical views. Cash in one AIM account is really no different than cash in another AIM account. It's all part of your wealth.
So if you are okay with using your cash to invest in an equity that is lagging (i.e. going down) as directed by AIM, you should also be okay with using your cash to purchase equities as designated by MPT.
Hi J,
Yes, you are correct in saying that ETFs are the equivalent of funds for the purpose of MPT. Otherwise they may not be.
"what would you think of letting AIM do the rebalancing for asset classes as the algorithm dictates instead of "by the calendar"?"
I'm not quite sure what you mean here, can you provide an example?
"my reading convinces me that asset diversity is crucial for optimal outcomes, there is SOME controversy about this. Are you firmly in the "convinced" camp?"
That depends on what you mean by "diversity." I believe that less correlation between securities in the portfolio will lead to better performance.
However even a portfolio that has relatively high positive (but not perfect) correlations between its component securities can benefit from MPT -- although not as much as if its components were random or negatively correlated.
I also believe that for MPT to give the greatest benefit, non-market risk must be diversified away. That's another reason I recommend MPT for use only with funds (most people don't have portfolios large enough to let them choose the proper individual stocks that will eliminate non-market risk).
What I am firmly convinced of is that diversification and proper asset allocation is absolutely necessary in order to wring out the maximum returns at the minimum risk. Unfortunately most people I speak to have different interpretations of what it means to diversify and how to do it properly.
It's similar to people who invest by the seat-of-their-pants with no plan. They know they should be investing, but they don't know how to do it properly -- and the term "investing" means different things to them.
So they simply buy the latest hot stock and hope for the best. Then when everyone is selling, they sell right along with them. Of course AIMers don't have that problem -- too much
So while AIM gives us an investment plan, it doesn't solve the asset allocation problem. Therefore there are AIMers (and I'm sure not all of them) that use a seat-of-the-pants method to diversify. They know they should be diversifying, but they don't know how to do it properly. In essence they don't have a plan when it comes to asset allocation so they go with a gut-feel approach.
They're essentially in the same boat as the seat-of-the-pants investor at the portfolio (macro) level, but they're far better off at the individual security (micro) level.
And that's where I believe MPT and other proven asset allocation methods come in. It removes the seat-of-the-pants asset allocation strategy and replaces it with a mechanical method based on sound reasoning and mathematics.
However just as blindly following AIM without any thought as to what goes into an AIM account can lead to disaster, blindly using MPT without any thought can also lead to disaster. MPT is simply a tool that, when used correctly, can provide significant benefits.
And Warren Buffett does it again...
http://www.fool.com/news/commentary/2003/commentary030813bm.htm
How's that for staying in the forefront?
Hi Tom,
Yes, you're right. The optimizer uses historical data to determine the means, semideviations and correlation coefficients. This works well for most mutual funds but not so well for individual stocks.
That's why the portfolio optimizer only uses MPT for funds and uses another method (based on William Sharpe's work) for individual stocks.
"How would rebalancing be represented in the next view? Would it contain the results of the previous optimized account or be theoretical to the new point in time?"
There are bodies of work dealing with both views. AI implements the latter method and creates a new (i.e. independent of the previous recommendation) asset allocation for each new optimization.
"How often would one go in and select to "Optimize" the same portfolio again?"
There are no hard and fast rules for rebalancing as it depends on your investment timeframe as well as the AIM managed securities' behaviour. However if you have a long-term investment horizon, a good rule of thumb is to rebalance every 2 years.
So every 2 years you'd re-run the optimizer to determine the new portfolio allocations and then remove cash/equity from the funds that are above their designated allocations and add that cash/equity to funds that are below.
On the other hand, if one of your funds suddenly takes off and now comprises 90% of your portfolio, you might decide to rebalance before the 2 year period has expired.
For those with shorter horizons, rebalancing annually or even quarterly might be more appropriate (since I do know some people use AI for short-term investing -- although I don't recommend this).
You do bring up a good point however. Periodic rebalancing is essential to getting the most from any diversification strategy.
Hi Conrad,
I believe that if an investor wants to transfer cash from one AIM portfolio to another, it should be a concious decision and not one that is automatically made by a deep diver.
By having a common cash reserve, you let the deep diver make the cash reserve allocation decisions rather than maintaining control yourself. This could lead to one portfolio generating lots of cash (because it is rising) while another is consuming lots of cash (because it is falling).
In that case when the first portfolio needs the cash again, it won't have it (because the second one has used it). So the first one suffers although it was being prudent by selling portions of its holdings to accumulate cash for a rainy day. The rainy day arrives and it discovers that its savings have been used up.
If, on the other hand, the investor decided that it was a good idea to move cash from the high-flier to the deep-diver (because of some fundamental reason, or whatever), then that's an entirely different scenario.
That's why I don't believe that common cash reserves are as good as separate ones.
Of course I'm speaking from a strictly logical cash reserve view. All cash would actually be held in one cash reserve at the brokerage account level, however it would be managed by individual AIM portfolios at the logical level.
Hi Patrick,
" It also changes the Profit/Loss figure by now showing XX$ profit, whilst it does also correctly increase cash."
Yes, when you use the add interest/dividends it adds to your profit. So you'd like a new function that adds cash, doesn't adjust portfolio control and adds the new cash to the original investment rather than the profit? I'll have to think about that for a bit and if it makes sense for the general AI user, I'll add it to an upcoming release.
"When I was setting these up (3 positions in one portfolio to show a common cash reserve amount rather than a per diem allocation) I'd transposed the price on the first fund that I entered, $15.72 instead of $15.27."
I see what you're saying. However if AI allowed you to change the price of your equity, this could affect Buy and Sell recommendations that were given after that checkpoint. In your case there were none (since you had just started the portfolio), however it's also possible that there could have been some recommendations. This is an example of an "unsafe" edit.
While I see the problem, I can't think of an easy way to get around it. The Undo function allows you to fix errors in a "safe" manner -- although it can take more work on the user's part. However it is better to do the additional work (especially since this type of thing isn't done frequently) than to risk an unsafe edit causing out-of-sync problems in a portfolio.
" had also tried to walk the portfolio individually through the end of July to bring it current to early August and I found a little 'bug' - as I was entering cash for a prior date, AI posted it in the running history using the current date, even though I'd given it the earlier date in the posting function itself."
One of the things you have to watch for is that AI automatically defaults the date to the current date. If you forget to change it to the desired past date, you'll get the current date. Are you sure you didn't forget to change the date? If you think you did everything correctly and there was still a problem, please send me the step-by-step details and I'll try to recreate it here.
"I also found that once there's a later or current date in there one can't post anything earlier."
Yes, that's working as designed so as to eliminate the "unsafe" edit problem.
"One more thing, on Windows 2000 Professional, when I exit out of AI if I bring up the task manager it still shows it as a running application."
Now that's interesting. I haven't heard of that problem before. I just tried it on one of my computers (also running W2K Professional) and it closes down correctly. No AI processes running after closing the program. Is this consistent on your machine?
One final note, I'd like to keep this board's focus on AI enhancements and new ideas -- rather than support related issues. You can send support questions to the AI support desk (see http://www.automaticinvestor.com/contactus.html for email address) where they will be promptly handled.
Of course new ideas, like your first suggestion, should definitely be posted to this board.
Hi Irwin,
The allocations when using the Bullish Model for all ETFs would be as follows:
IBB 36%
IYC 2%
IYE 1%
IYH 6%
IYW 55%
Note that this is an Alpha version of the Portfolio Optimizer, so please don't use these values for any real portfolios. They are for demonstration purposes only.
Before it's released for public consumption I run a variety of tests to ensure all of the underlying calculations are correct.
For example, I'll perform the correlation coefficients' calculations using Minitab (a statistical software package I have) and check that they match the calculations in the Portfolio Optimizer.
As these tests haven't been done yet, it's not a good idea to use these data for real portfolios.
Let me know if you have any suggestions.
Hi LC/Jibes,
This is very interesting. About 6 months ago I also began experimenting with just such a system. However the "factor" is based on the Fibonacci sequence.
I implemented it as a software program and ran some backtests that had promising results. However it still needs tweaking. I haven't had a chance to add the tweaks yet as I've been busy with quite a few other projects.
It's good to know others have independently come up with similar systems. That gives me even more confidence in it and just might push it up my priority list a bit.
I'll look forward to seeing your respective systems.
Hi Patrick,
Here are some thoughts on your comments.
"Currently my positions have me at 12% cash reserve. I'd like to increase this to 20%"
You should be able to do what you want by using the "Add Interest or Dividends" function on the "Cash Reserve" menu. This just adds cash to the account without adjusting the Portfolio Control or requiring you to purchase equity.
"Another feature I'd like to see in AI is the ability to edit a transaction after it posts into the system"
This is a feature that other AI users have requested. The problem is that some of the changes might affect AI's recommendations from that point on. In those cases, it wouldn't be possible to follow the new recommendations because they would be in the past. Therefore things could get quickly out of sync with reality.
If you have some examples of the kind of edits you're thinking of, I will look at them to see whether they're "safe" edits.
"is it possible to set things up in the software to where one can have individual portfolios but a common cash account that feeds all of them?"
I'm not 100% sure what you mean here, but I'll take a guess. If I interpret it one way, here's the answer...
1) You can view AI portfolios as a logical representation of the underlying brokerage account. Therefore you might have multiple AI portfolios each holding a different security and each with its own cash reserve. However all of the securities and the entire cash reserve could be in one brokerage account.
AI manages each portfolio separately and doesn't know about how the securities are physically held.
Another interpretation is that you want to have individual AI portfolios all sharing one cash reserve. If so, here's the answer...
2) First, I don't think this is a good idea because securities that fall dramatically can eat up the cash reserve and cause other security portfolios to run out of cash (through no fault of their own).
However if you still want to do this, you can simply withdraw cash from one or more AI portfolios (and reset its/their PC(s) using the "Adjust Portfolio Control" function on the "Tools" menu) and deposit it into the portfolio that requires it.
If everything is in one underlying brokerage account, nothing needs to be done at that level.
I hope I've answered your questions. Let me know if you have any others and please send me some examples of your "edit" request so I can get a better idea of what you have in mind.
I'm always looking for user feedback in order to keep AI at the forefront of AIM software, so your suggestions and comments are especially welcome.
Yes, it was Triple "O," you remember well.
Hi Conrad,
"This A.I. optimization looks like a piece of cake..."
Well using it certainly is a piece of cake and sticking to the recipe will most likely give you a bigger piece of the pie. I guess you can have your cake and eat it too.
Speaking of pies, I'm not sure if you ever went to White Spot while you were in BC, but they have a blueberry pie that's the BEST blueberry pie I've ever had. No top crust, just a bottom crust filled with prime blueberries and a special glaze.
A dollop of real whipping cream and there's not much more you can ask for
Automatic Investor 3.0 is scheduled for release later this year.
It will include the Genetic Parameter Optimizer, that I mentioned a little while ago, as well as the Portfolio Optimizer that will provide systematic Asset Allocation capabilities to AIM portfolios.
A few people already have the Genetic Optimizer in its Beta incarnation so I won't say anything more about that, however the Portfolio Optimizer is something that I've wanted to add since the AIM 2001 convention in Las Vegas. Unfortunately there were other more pressing enhancements that had to be made. But now, it's finally implemented and working.
It uses Harry Markowitz's Modern Portfolio Theory (for optimizing Mutual Fund portfolios) and a heuristic based on William Sharpe's Sharpe Ratio (for optimizing individual stock portfolios).
An improvement has been made to use the semideviation (or downside risk) rather than the standard deviation. Markowitz has said that semideviation is a better way to go, but for some reason he stuck with standard deviation as his risk measure.
After a few weeks of testing I believe that semideviation is the superior way to measure risk -- so that's what Automatic Investor uses.
Here are some screen shots that show the AI Portfolio Optimizer...
The Optimizer screen contains three tabs.
The first tab is where you define your portfolio. You choose the securities of interest and select the Model that will be used to manage each respective security with AIM (note that the standard AI Conservative, Moderate and Bullish models correspond to Lichello's AIM settings as described in the 1st, 3rd and 4th editions of his book respectively).
Once your security symbols have been entered, you click the Optimize button to start the optimization process. Optimizing is a relatively fast process (even with many securities) and ends by choosing the best portfolio out of all the possibilities it has considered. It then displays the result in two views.
The first view can be seen using the second tab as shown below.
It provides a pie chart of the recommended asset allocation.
The second view can be seen using the third tab.
It provides a report that contains the recommended asset allocation as well as the Models used to optimize the portfolio.
As I previously mentioned, AIM portfolio optimization *must* use the AIM results -- not the underlying securities' price movements. This means that all semideviation and correlation data must be calculated from the AIM results. This can now be easily done within AI.
You'll notice that portfolio optimization and asset allocation is now a very simple task. However although it is easy to do with AI, the underlying theories and calculations are quite complex and up until the advent of the PC required very specialized and expensive computing facilities. MPT even won Nobel Prizes in Economics for Markowitz and Sharpe.
Technology in the hands of the masses is truly wonderous indeed.
As usual, if there are any questions or suggestions, please feel free to send me an email.
Hi Conrad,
You are right in that the different quote services all have different formats. However there are two things you can do.
First, some quote services (such as nasdaq.com) have an XML API that returns quotes as XML documents. This allows you to use an XML parser to retrieve the relevant information.
Of course you still need to know the specific XML vocabulary that particular quote provider is using, but it does make it easier. Unfortunately not too many services are currently doing this, however I believe it will be done more and more in the future.
Another, related, technology is Web Services. If you want your software to be easily maintained, then Web Services will be the big thing in the near future. Again, unfortunately, Web Services are in their infancy, so you're probably out of luck on that front.
So where does that leave you? Well, it leaves you with option number 2, having to determine the quote service's format and writing a parser that will extract each piece of datum (such as high, low, closing price and such).
AI uses Yahoo! Finance as its quote source. Yahoo! allows access to foreign exchanges by adding the appropriate suffix (see http://help.yahoo.com/help/us/fin/quote/quote-19.html for a list of suffixes).
This should be good enough for you to interface with most major European exchanges and I'd recommend this as the way to go.
One note on writing a parser. There are good ways to write parsers and bad ways. Bad ways make them very specific to the quote service provider. This is *bad* because not only can you not use other quote services easily (without having to rewrite the parser), but if the quote service changes its format, you have to rewrite your parser (Yahoo! Finance has changed formats a few times over the past 3 years).
The good way is to write a general parser that allows you to specify configuration information (using a text file, XML file, database fields, etc.) for the data items as well as their relative positions.
This means that if something changes (or you want to use a new quote source) you simply change the configuration information rather than having to rewrite the parser. FYI, this is what AI does.
I hope that helps. Good luck!
And Hello yet again,
"Now, I need to learn what he's talking about (mean-variance optimization, or somesuch)."
You can get a good overview here --> http://www.automaticinvestor.com/articles/mpt.html and some other ideas here --> http://www.investorshub.com/boards/read_msg.asp?message_id=1230900
Hi again Jack,
FYI, I've implemented a genetic algorithm in Automatic Investor 2.0 that finds the best (well, almost best) set of AIM parameters. I've added it to the brute-force algorithm that finds the *best* set of parameters (however it can take an unreasonably long time to do so).
This sounds similar to your evolutionary programming technique (although it's not identical).
I've also used a genetic algorithm to implement an MPT-based portfolio optimizer (based on work by Markowitz and Sharpe).
Seems to work very well.
Hi Jack!
Welcome back. It's been a long time. I don't know if you remember, but I had corresponded with you about 3 years ago about some of the work you were doing at thinkalong.com.
After all this time I still find it very interesting and one of your enhancements (Control Increment, remember that?) made it into Automatic Investor 2.0.
I hope you have time to share your latest and greatest ideas and innovations.
Hi Rick,
Yes, ETFs (and index funds in general) can have large portions of their value attributed to one or two of their components (especially when those components have risen rapidly).
However IBB is still diversified and holds about 75 individual stocks. This is something that the average investor could not (or would not want to) do. So by purchasing something like IBB, you minimize non-market risk in that sector. And since most of the holdings are highly correlated with one another, you don't sacrifice the usual volatility needed for AIM.
Once you start diversifying over a number of different sectors, you'll actually start to minimize non-market risk over the entire market (and again, you're not giving up the volatility).
" Wouldn't it be better to pick 4 stocks in each group and use their volatility to Aim them?"
You could certainly do this, however if you have 10 sectors, that's 40 stocks you have to look after. In my opinion, that's far too many unless you invest full-time. Furthermore, you're not as diversified as you would be in an ETF. So you open yourself to more risk.
I think that ETFs are a good compromise in returning reasonable reward for risk. On the other hand, investing in indvidual stocks can be more lucrative, however there is also more risk involved.
Hi Tom,
"Is there a way to figure the individual equity/cash ratios into the equation?"
I think you could probably do it, but it would be quite complicated and I'm not sure you'd gain anything.
My idea for using AIM with MPT is to use the AIM returns and AIM standard deviation (rather than those of the underlying security) and feed those numbers into the mean-variance optimizer (so the equity/cash in the individual AIM accounts are already taken into account before MPT enters the picture).
Then you'll have a portfolio consisting of various weights for each security (e.g. 15% to security A, 25% to B, 20% to C, 30% D and 10% to E). You then invest the recommended number of dollars by starting an AIM account for each of the respective securities.
After that, you simply manage each AIM account as usual until your rebalancing period comes up. Then you do the same thing to determine the new allocations. Once you have the new allocations, you remove funds from those that are over their allocations and add those funds to accounts that are under their allocations.
So basically you would do what you're currently doing, but rebalance periodically as directed by the mean-variance optimizer.
One interesting thing you brought up is using the IW to determine asset allocation. That's using the IW at the macro level (in addition to using it at the micro, AIM, level). Sounds like something I can think about.
"Who would have guessed last September that the two best performers in the portfolio would be Tech and Biotech?"
That's the beauty of diversifying properly, we don't have to guess
Hi Tom,
What you're doing is almost exactly what I was talking about back at the last AIM conference. Basically you've let AIM manage your ETFs at the micro-level and you've let diversification manage your overall portfolio at the macro-level.
The only difference is that my suggestion was to use the MPT algorithm (i.e. a mean-variance optimizer) to optimize the portfolio and determine how much of your total portfolio to invest in each security (ETF in this case). How did you decide how much to allocate to each ETF?
Although I hadn't tried it in real-life, my theory was that an MPT diversified portfolio would exhibit much less volatility at the macro-level even if the micro-level investments were very volatile. And it looks like your results are in line with that.
I'm still an individual stock type of guy, but I'm actually considering moving a portion of my investments to ETFs at some point in the near future (so I can use a mean-variance optimizer to determine optimal allocations. For individual stocks I have another method of determining allocations).
As usual you're at the forefront of implementing good ideas in AIM. I too will be following your ETF portfolio with interest.
Thanks for the update!
Hi Irwin,
"how many ten baggers does someone need?"
42
http://www.bbc.co.uk/dna/h2g2/alabaster/A846434
Hi Rien,
"While I also must admit that when you mention "expected return" the hairs on the back of my head start tingling a bit. "
Then your Spider-sense is working well You're right. Estimating Expected returns for use with MPT is a HUGE problem when using MPT as a predictive device.
That's why I don't suggest it be used for individual stocks (see my post that started this diversification thread at http://www.investorshub.com/boards/read_msg.asp?message_id=1230900 ).
Rather it should be used with index funds (including sector funds as used by Jennie, Tom Veale and others) and other assets where returns can be confidently predicted.
Hi Rick,
"Why does it have to be 1/2 of the value after you buy or sell"
Actually it's only 1/2 the value after you buy. Portfolio Control is not modified after a sale. However you'd like the reason for this...
And the reason is... because Lichello said so. That's it really. Lichello experimented with other values and came to the conclusion that 1/2 was best.
Keep in mind that Lichello did all of his testing by hand and didn't have access to a computer. So he was severely limited in what he could test.
Today, of course, most of us have computers and we can run all sorts of tests. If you do run these tests you'll discover that just as the 10/10 Split Safe settings aren't always ideal for all securities, neither is adding 50% to the Portfolio Control always ideal.
Sometimes it's better to add more, sometimes less (but generally 50% works fine). However this gets into AIM parameter tuning (or Model building as I call it) and that is an entire subject in itself.
"and then you can use split safe with that figure?"
I'm not quite sure what you're getting at here. If you elaborate, I'll see if I can answer your question.
Hi Rien,
"But if you rebalance periodically this chance increases back up to 1%!. This is because the rebalancing will eventually put all your money into the one asset class that goes to zero."
While this is certainly possible, it is not probable if you're diversifying properly.
The reason is because as the poor performing asset goes to zero over time, its expected return will be decreasing. Therefore when you rebalance (e.g. using MPT), this will be taken into account and that asset will most likely leave the optimal portfolio (i.e. there will most likely be another portfolio, that doesn't contain that asset, that will have a smaller risk for the same return or a greater return for the same risk of the best portfolio that contains the poor performer).
If you use semideviation rather than standard deviation as the risk measure, this effect will be more pronounced.
So diversifying and rebalancing would most likely result in you selling out of the poor performer well before it hit zero.
If the poor performer went to zero very quickly then, of course, you wouldn't have time to rebalance it out of your portfolio and you'd lose your investment in that security.
However diversification would *most likely* save you from losing all of your money because you'd still *most likely* have the value of the other stocks (the ones that *most likely* didn't go to zero) in your portfolio.
Hi Bernie,
"If you ask ten people to shuffle a deck of cards you will see that there is more than one way to shuffle cards. If you ask those ten people what they are doing they would all say that they were shuffling cards."
True, however if you defined a set of steps to shuffle cards (such as take the deck in your right hand and cut it in half, then shuffle the top half three times, then shuffle the bottom half two times, then put the bottom half on top of the top half, etc., etc., etc.) and gave that set of steps a name, such as the "Bernie Shuffle," then you can say that someone doing the Bernie Shuffle is shuffling cards.
However someone shuffling cards in another way is still shuffling cards, but he's not doing the Bernie Shuffle.
Similarly, when I speak about proper diversification, I'm referring to a specific set of steps (get your securities, choose ones that will minimize non-market risk, choose ones that have low correlations, etc., etc., etc.) I've given that set of steps the name, "Proper Diversification." So you can say that someone doing "Proper Diversification" is diversifying.
However although someone who throws equal amounts of money at a set of randomly chosen stocks is still diversifying, he's not doing "Proper Diversification" in the same way someone randomly shuffling cards isn't doing the Bernie Shuffle.
Proper Diversification is a subset of Diversification, so anyone doing proper diversification is diversifying. However the reverse isn't necessarily true.
Your point is playing on the semantics of the word "diversify," however that line of argument takes us off track and doesn't apply to what I was trying to say.
"Sorry, but I have to disagree with you."
I hope my explanation clarifies things, however if you still disagree that's okay. I'm not too interested in arguing semantics at the present time as I'd rather concentrate on the theory of Proper Diversification with AIM.
I know that you've been investing for quite some time and what you do is working out well for you. However I think there are others that are interested in diversification theories that are more specific than what you believe to be diversification.
Hi Tom,
Yes, that's actually how MPT was/is best used -- diversifying over different asset classes (bonds, stocks, etc.)
I've been concentrating solely on stocks, but your method is the better way to go if your portfolio is large enough.
Hi Bernie,
"Actually diversification has nothing to do with any investment method."
Yes, I agree with this. Although some investment methods make it easier to diversify than others.
"All one has to do is look in good old Webster's."
Unfortunately the dictionary's definition, although it can be correct, is too broad for what we're discussing.
Its definition allows us to simply buy various kinds of securities. However this isn't proper diversification. Proper diversification means that we need to purchase securities that not only eliminate non-market risk, but also have low correlations between them.
Then we have to buy specific amounts of each security. That's a big difference to simply distributing "investments among kinds of securities, or the like."
"It's amazing how the simplest concept can become so complicated."
The high-level concept is simple, however the implementation (as well as the low-level details) is anything but.
Hi J,
"If I understand the academic studies properly, risk is lowered AND returns are slightly better."
Yes, that can be the case. MPT tries to create portfolios that maximize returns for a given risk level or minimize risk for a given return (basically the same thing viewed from two different perspectives).
So if you have a non-diversified portfolio, that has risk X and returns Y, then it's quite possible that a properly diversified portfolio can give you X' and Y' where X' < X and Y' > Y.
"It seems counter-intuitive that a more conservative scheme produces better results but that seems to be the case."
Very well said. Much like quantum mechanics, this concept is not naturally grasped by our minds. It appears to go against common sense. However it has been proven over and over again to hold true.
Hi Rien,
"Have you (or anyone here) made this kind of return with AIM? I.e. >= 12% over a time frame of 10+ years."
I haven't been a long-term AIMer (only about 3 years) and I don't use AIM By the Book -- I use a modified version of AIM. So I can't comment on real world actual long-term results.
However I've averaged significantly more than 12% per year over the past 3 years (although much of that has come in the past year, but then that's the reward for buying all those relatively cheap shares when the prices were down).
On the other hand, I've run so many backtests in AIM that I'm quite confident of AIM's long-term probable results. If I didn't think AIM would be able to return significantly more than 12% per year, I'd simply invest in a good small-cap index fund.
"I would very much like to have the board's consensus on this."
Yes, this would be interesting to see. There are a number of people who have been investing with AIM far longer than I have, and I too would be interested in seeing their real world results.
"(Though I must also add that my LD-AIM silver stocks have probably made about 25-30% in the last two months)."
Congratulations! So you are somewhat diversified after all
Hi UT,
"Diversification is really just another way of saying lowering risk."
Yup, that's my thought in a nutshell.
Hi TF,
"For someone just starting out, who doesn't have a lot of money but will have more to invest over time, it is possible to use TIME to diversify."
Yes, that's a good point. As your time horizon increases, your risk level decreases. And as your portfolio value grows over time, you can then diversify properly via different equities.
"Personally I like "Exchanged Traded Funds" and the offerings by I-shares in particular."
I'm in total agreement here. One of the posters on this board introduced me to ETFs a couple of years ago and since then I've become more and more of an ETF fan (although I currently don't own any).
Tom Veale has information on this at http://www.aim-users.com/etfunds.htm
If you don't want to spend the time messing about with individual stocks, this is the way to go.
On the other hand, I'm completely against actively managed mutual funds (the next AI eGazette will have an article explaining why). I feel these funds are the biggest rip-off currently going on on Wall Street.
Hi Rien,
"when starting with $10,000 or even $100,000 it is next to impossible to make it to a million when one uses diversification."
If you start with $100,000 then you can be almost guaranteed to make $1 million in 21 years if you have AIM manage a well-diversified set of stocks returning 12% per year. However I'd suggest that most AIMers can do better than 12% per year.
Starting with $10,000 will take a longer period of time. However I'd say it would be almost impossible to make it to $1 million (starting from $10,000) by not diversifying. The risks you'd have to take would be quite high, or the timeframe would have to be prohibitively long for the "safer" investments.
Of course, with $10,000, you'd have to start out without much (if any) diversification, because $10,000 doesn't allow you to properly diversify. However as your investment grows, you should then start to diversify.
At the individual stock level AIM manages your risk by moving you in and out of cash. At the portfolio level, properly diversifying manages your risk by reducing (or eliminating) non-market risk.
I don't think we can solely look at returns. Returns have to be looked at within the context of risk too.
Further to my previous post on diversification, I've also been looking at semivariance (a measure of downside risk). The Sharpe Ratio uses the standard deviation in its calculation. However StdDev doesn't differentiate between variances above the mean and variances below it.
Of course in investing, we don't mind huge variances above the expected return (i.e. no investor in his right mind would classify an abnormally high positive return as a risk, however most investors would classify a negative return as a risk).
Semivariance also works when returns aren't normally distributed, whereas StdDev doesn't.
I've been testing Semivariance-based diversification and it appears to provide much better results than StdDev-based diversification.
Interestingly enough, semivariance is a specific case of something called the Lower Partial Moment (LPM). Using general LPM investors can choose how risk-averse they want to be.
I'll probably provide more details at some point on the Advanced Automatic Investor board (http://www.investorshub.com/boards/board.asp?board_id=1172 ) once I've had a chance to run more tests.
Hey Rien,
"If you are in the market to GET rich, then diversification is a killer."
I look at it from a slightly different perspective. If you're trying to get rich, you still need to diversify or else you're exposing yourself to risks that, in all probability, will cause you to lose money over the long-term and inhibit you from getting rich.
The slow, steady and consistent building of wealth gets you rich. On the other hand, if you're trying to get rich quickly, then I would agree that diversification is a killer. However, even then, most people who try to get rich quickly fail.
The only time I would agree with not diversifying is if someone is starting out and doesn't have enough investment funds to properly diversify. At that point it doesn't make much sense to diversify because the commissions would become a significant expense. In that case you invest your funds in a good stock and hope for the best (or invest in a good index fund, but that's another story).
Hi UT,
Yes, not everyone thinks diversification is a good idea. Some of that stems from the fact that people use the term in a variety of *different* ways.
Mutual funds, for example, are said to be diversified. However as far as I'm concerned, most are over-diversified (especially the large ones) which is almost as bad as not being diversified.
I mean large funds have to be. When you're managing that much money there are only so many companies you can hold before you need to reach down to the second tier, and then the third tier, fourth, fifth and so on. That's not proper diversification.
"The way to limit risk is to limit loss."
That's true, but I don't believe that a stop is the way to do it for an investor (although I completely believe that *traders* should use them).
What happens is that you get stopped out of the stock when the price temporarily goes south. Using AIM and a stop loss doesn't make sense. Using a stop when you have a long term investment horizon also doesn't make sense.
"Diversification also comes with the understanding that you need many stocks to diversify."
I'm not sure what you mean by many, but I think 10 or less should do it. That's not too many in my opinion.
"The first week I posted on this site I mentioned that Warren Buffet was against diversification and everyone lambasted me as being insane."
Well, not everyone. I didn't
"Diversification is a protection against ignorance. (It) makes very little sense for those who know what they're doing."
First, I think Buffett is using the term in a different way than you think. I mean look at BRKA. It's a conglomerate with interests in fast food, insurance, finance and more. That's as properly diversified a portfolio as I've seen.
However Buffett didn't just go out and pick up a host of companies in the name of diversification. He built his holdings by choosing great, undervalued companies first and then ended up being diversified.
On the other hand, even if Buffett meant he truly doesn't need to be diversified, I believe him. He's Warren Buffett. But that doesn't mean the average (or even above average, actually let's just say everyone except Buffett and a handful of his peers) investor doesn't need to be.
If you follow hockey, you'll know that Bobby Orr was the greatest defenceman to ever play the game. Before Bobby, defencemen never made forays into the offensive zone. Rather they stayed back on defence.
However Orr changed all that by making end to end rushes and scoring more goals than many of the forwards of the era. He was able to do this because he was an exceptional skater and could anticipate when he could go and when he should stay back.
His skating ability and speed allowed him to get back to his defensive position even when he lost the puck in the offensive zone.
At the other end of the spectrum was a defenceman named Dave Babych. Even if you're a hockey fan, you might not have heard of Dave Babych (he played for the Vancouver Canucks).
The reason is that he was as slow as molasses and watching him was anything but exciting. If he decided to make a rush into the other zone (which he never did), I'm sure he'd lose the puck, but if by some miracle he didn't lose it and garnered a shot on net, he wouldn't be able to get back into position quickly enough to do his job. That's the difference between Bobby and Dave.
And while most of us would like to think we're the investment equivalents of Bobby Orr, we're usually really a bunch of Dave Babychs. Whereas a guy like Buffett is a Bobby Orr. So if he thinks he doesn't need to diversify, perhaps he can get away with it. Mere Dave Babychs can't.
"In my opinion, diversification for diversifications sake is a surefire way to not lose as much money, but to slaughter your potential gains. I currently hold 20 stocks..."
If you're saying what I think you're saying here, then I agree. Diversification for the sake of diversification is bad. However proper diversification is not bad. Not diversifying is bad.
The fact that you own 20 stocks (and I'm assuming not all of them were purchased with the aim of diversification) means that you're already diversified. Why not try to determine how to optimally diversify your portfolio with those same stocks? I mean you already have some diversification going there, you might as well make it better.
"This further enforces my view that I should have chosen less stocks."
I'd agree with you that you would be better off with less than 20 stocks.
"put more money in my 'sure bets'"
Now that's the ultimate question. How do you know it's a "sure bet?" And if you do know, then why not put all your money in it rather than just "more" money in it?
My guess is that it's not a "sure bet," rather it's a solid stock with a low probability of going bankrupt. However there's always a chance. To make a long story short (too late, I know that's why we need to diversify.
"In any event, just my opinion"
I always enjoy reading your opinion.
Hi Steve,
"How would one go about selecting the candidates to start?"
I would select candidates based first on the criteria I like. These include, but aren't limited to, Earnings Growth, Cash Flow, Market Cap., debt, and volatility. I would be happy to invest in any company that makes this list.
Next I would sift through these and choose the ones that aren't in the same sectors. I'd also look for stocks that aren't highly correlated. From time to time I might also use the Sharpe Ratio to help me decide between two stocks that I like equally.
Finally I'd select the ones in which I will actually invest. As I mentioned previously (http://www.investorshub.com/boards/read_msg.asp?message_id=983734 ), I would limit myself to, at most, 10 stocks (currently I have 7 in my main portfolio -- not including the ones I speculate with and which are purely short-term holdings).
"Are there any actual results one could point to with respect to an actual set of holdings that were balanced using MPT and then let run? How did these results compare to the equal mix or other methods?"
There are no results for AIM, however there are examples all over the Internet that show diversification leads to significantly better results. However I've not found an exhaustive study based on strict statistical testing procedures.
Having said that, I've been looking at quantifying diversification for a couple of years now and am fully convinced that investors *must* do it if they want to limit their risk *and* increase their long-term returns.
I'm currently working on putting together a testing suite for AIM that will allow me to run proper backtests on the effects of systematic diversification. This will allow me to test the different diversification methods and obtain empirical evidence on their relative performance.
"Knowing your time zone, and noting the time of your post; I'm compelled to ask. Did you stay up all night to compose it?"
Actually I do much of my investment work late at night (I've been a night owl for as long as I can remember). Yesterday I was writing a portfolio optimizer and was just about to head off to bed when I decided to write that post.
I figured it might help someone since most people know diversification is good, but most don't know how to go about properly doing it.
From your previous posts I see that your investments are doing quite well so far this year. Congratulations!
The most important concept in investing is diversification.
Regardless of the actual strategy we use, if we don't properly diversify we open ourselves to more risk than we need to take.
The question, of course, is how do we properly diversify?
There are a number of strategies that range from completely useless to very good. I'll mention a few here.
EQUAL FUNDS
The simplest way to diversify is to decide which stocks you'd like to own and divide your investment funds equally between them. Then, periodically, you rebalance. This diversification strategy is better than nothing, but lacks any real ability to use synergies that might be present in your portfolio.
SHARPE RATIO
A better method is to use the Sharpe ratio (SR). Since you can type "sharpe ratio" into Google and read all about it, I'll refrain from giving the details here. Suffice it to say, however, that it is a measure of reward/risk. The higher the value, the better the investment (from a reward/risk perspective).
Note that a high sharpe ratio doesn't necessarily mean the highest return or the lowest risk. Rather it means that that's the best you could have done when reward and risk are taken together.
To calculate SR, for a stock, you need the stock's expected return and standard deviation. You also need the best risk-free rate (typically what they're paying for cash).
For AIMers, you need to calculate the return and standard deviation on AIM's results, not the underlying security's price.
This makes it a little more difficult to use because you have to obtain your AIM results and then use them to calculate expected return and standard deviation. If you use the underlying security price directly, it won't work correctly for AIM.
Okay, so we can calculate the SR. Great! Now what?
As it turns out, once we have the SR for all of our stocks, we can use a simple method to diversify more intelligently than just throwing equal amounts of cash at our stocks.
Here's what we can do...
1) Compute the SR for each stock in our portfolio.
2) Sum all of the SR values (I'll call this the srSUM).
3) Divide each stock's SR by srSUM. This gives the percentage allocation for that stock.
For example, let's say I have 3 stocks in my portfolio (A,B and C). Let's further say that I've calculated the SR for each stock as follows:
A has an SR of 2.5
B has an SR of 1.5
C has an SR of 4.0
In this case srSUM is 8 (i.e. 2.5 + 1.5 + 4.0).
I then divide the SR for stock A by the srSUM of 8 to get 2.5/8 = 0.3125
If I do the same for B and C, I get 0.1875 and 0.5 respectively.
This means that I should invest about 31% of my money in stock A, about 19% in stock B and about 50% in stock C.
The advantage of this strategy is that the better reward/risk stocks get a larger share of our investment funds (which is a far more logical way to go than the simple equal allocation strategy I described above).
However this strategy doesn't take into account the possible interactions between securities in a portfolio. But these interactions can be significant (for example, adding a risky stock to a less-risky portfolio can actually cause the entire portfolio's risk to decrease).
MODERN PORTFOLIO THEORY
A strategy that does take interaction into account is Modern Portfolio Theory (MPT). This strategy seeks to provide portfolios, consisting of various weightings of stocks, that maximize returns for a given risk or minimize risk for a given return.
Calculating these portfolios require you to have the Mean and standard deviation for a stock as well as a covariance matrix (which is just a table that lists how stocks behave relative to one another). You can also get by with correlation coefficients (which you can then use to calculate the covariance).
For those that don't like to think about these things, the reason I bring it up is to explain a very important point.
First, you have to either estimate or otherwise obtain (e.g. via historical data) these values in addition to estimating or obtaining the Mean and standard deviation values.
If you happen to be off by just little (especially for the Means), the resulting asset allocation can be significantly changed. It's hard enough to obtain a reasonable Mean, but adding covariance values (you need one between each pair of stocks) just increases the odds that you won't get something quite right.
Once you have these values you can feed them into an algorithm, called a Mean-Variance optimizer (or you can use some other algorithm that does almost the same thing), that will calculate your optimal portfolios.
You can then select a portfolio based on either the maximum risk you're willing to assume or the minimum return you're willing to accept. Once you've selected your portfolio, you invest your funds according to the specified allocation.
SHARPE RATIO AGAIN...
Interestingly enough, you can use the Sharpe Ratio to automatically select the best portfolio for you (from a reward/risk perspective). This frees you from worrying about what risk you're willing to assume or what minimum return you'd like. Once MPT has returned the set of efficient portfolios, you calculate the SR for each one and choose the one with the highest SR. It's easy and it's automatic. Your computer can even do it for you.
Note that, as above, you need to calculate the Mean, Standard Deviation and Covariance on your AIM results -- not the underlying security.
Unfortunately, because of the estimating inputs problem, MPT doesn't always work well in the real world. For historical data you can't beat it. However as a predictive device, it can fall short.
Having said that, however, MPT works fairly well if you can somehow feed it accurate inputs or you use index funds over long periods of time.
THAT'S ALL THERE IS TO IT
So there you have it. Three methods of diversifying your portfolio. If you're not currently diversified, you should be. Even using the "blindly allocate an equal amount of funds to a number of stocks" method is better than nothing.
However if you invest just a bit more time, you can use the work of a couple of Nobel Laureates to more intelligently diversify your portfolio. And that's got to be worth a huge sum of money somewhere down the road.