Saturday, 24 February 2007

Predicting the Market

No matter whether you've been investing for a day or a decade, there's always the niggling thought that someone out there knows the "secret" to timing the market. So you may give your money to a professional to manage (eg. Actively Managed Funds), or read books or take courses on a technique to boost your returns above the market average. All these things will cost you time and/or money. Occasionally one method you try for a while (day/month/year) will actually produce good results, so you'll tend to think you've discovered the "secret" and stick to that method until it ultimately fails (and in the meantime possibly blog about it, tell your friends, write a book, or teach your "secret" in seminars).

I must admit that I've done this over the years, and I still will invest a portion of my money using particular techniques in attempt to boost returns - for example my "Little Book that Beats the Market" US Stock Portfolio (Value Investing), and toying with the idea of moving some of my domestic stock investments from my geared personal stock portfolio (stock picking) into a professionally managed individual stock account (with Direct Portfolio) within a self-managed super fund (SMSF). But my preference over time has shifted towards low-cost index funds where available (I'm thinking of moving my retirement account from my employer's default fund (Westpac/BT Employer Super) into a SMSF where I can invest in Vanguard Index Funds or ETF such as the Commonwealth Diversified Share Fund (CDF)

One of the main reasons I've drifted towards a preference for Index Funds is that although I recognise that some actively managed funds outperform for extended periods (eg. Berskshire Hathaway), most funds that outperform for a while (up to a decade) can be simply put down to "luck" or random chance. Similary, techniques such as charting, while always able to explain stock movements in hindsight, don't appear to have any real predictive power. If you ever need reminding that a LARGE component of stock price variations (and market variations) is purely random, despite the appearance of clear patterns or "trends" in the charts, just do a simple random walk simulation in excel.

For example, just as a reminder to myself, I ran a simulation in excel to model the coming year in the ASX All Ords using a few basic parameters and a random number generator:

Simulation Period: Daily Index Value for next 250 days
Starting Value: 5800 (around the current Australian market level)
daily movement formula: new value=P+(10%/250)*P+RAND()*200-100
where P=previous cell's value (eg. B4, if calculating value for cell B5)
I picked an overall 10% pa ROI as a typical stock market trend and a daily random move of + or - up to 100 points as a fairly "typical" market movement.

It's amazing how realistic the "chart" for this simulated market is every time you press F9 to recalculate the random numbers. For example, just from random numbers, you can get a continuing "bull market", a sudden "crash", a "bear market", or a "correction":
Bear Market:

Bull Market:


Crash: (just a little one)

Of course the market (and individual stock prices) isn't purely random - but key events that will shift the market in a particular direction by a significant amount aren't known in advance (otherwise the effect would already reflected in the price by other investors trades adjusting the price level).

Ultimately, I think the time spent trying to uncover "secret" techniques to predict the market would be better spent by a novice investor in a second job earning more funds to invest, and for a more seasoned investor with a significant amount already accumulated, just concentrate on diversification, asset allocation, tax-effective investment structures and try to avoid high fees, churning or other return-diminishing behaviours.


mOOm said...

Long segments of daily or weekly closing stock prices pretty closely approximate a random walk (with drift) at a first approximation. But there is subtle structure in the short term underlying stock price movements which can be recovered with high levels of statistical significance. This isn't the conventional view. But what I have been working on in my model development in the last year (and in earlier more primitive models). Now I just need to demonstrate it with consistent profits :)

But I do have training in applied time-series econometrics. Most trading systems I read about don't sound a lot better than random - the edge is subtle and it is all in the money management techniques.

The Bull said...

Interesting charts there. I have replicated them myself in Excel and had a play with them.

I am curious as to why you wish to establish a SMSF and invest in index funds, rather than just invest within index funds within your current Mastertrust (or an alternative mastertrust). Would the costs of administering a SMSF not erode the cost savings you would make by investing in index funds? said...

Good Luck to moom - I've read some research that suggests there are some slight patterns that are able to be identified in short term stock price movements, but generally their magnitude is less than the trading costs involved. If you can find some patterns that are significant enough to profit from trading, you should keep it secret, borrow a large stake and make your m(b)illions - just hope any historic patterns continue as long as you are trading them. Remember that sudden catastrophic effects can overwhelm subtle trading patterns - eg. Long Term Capital management.

The Bull - I currently have around $340K in super with the BT Employer fund that my employer usually uses. We get a "rebate" of fees, but it still has an overall admin fee of around 0.7% after the rebate, and then you have to add on the management fees of whichever investment funds you are in (usually around 1.2-2% for AU and OS stock funds). There's no option to invest in Index Funds within the BT scheme.

I'm thinking of setting up a SMSF using It costs $600 pa for admin/audit/reporting, and no setup fee. There's another $150 or so payable to the ATO for a SMSF - but this still adds up to a total of only $750pa, which equates to 0.22% on my current super balance, and will be a smaller % as my super builds up over time. Once in the SMSF I can invest in similar managed funds as I currently hold, or invest in Vanguard Index Funds (around 0.5% - 0.75% MER), or invest in shares I select myself, invest in an individual actively managed stock portfolio run by, or whatever legal superannuation investments I choose. I can also have my wife's and sons super in the same SMSF which will save further on admin fees. I think moving my super to a MasterTrust would cost more than this version of a SMSF.

The major downside of moving to a SMSF for me would be the hassle of arranging replacement life insurance (as my insurance is currently via the BT super fund), which might cost more, and may also involve a medical for the amount of coverage I currently have. I may look into moving the bulk (say $300K) of my super to a SMSF and keeping a small balance ($40K) in the BT Employer fund to keep the insurance coverage.

If I setup a SMSF I'll probably move my other stock investments within the scheme to save on CGT when the investments are sold (there is apparently 0% CGT rate if super assets are sold when in the "pension" phase ie. once I'm retired and over 60).

I'll be doing some more research, calculating CGT liability if I sell off my current investments outside super to shift into a SMSF, and so on, prior to July 1 this year. I don't have enough to invest in super to worry about the 30 June cutoff for investing $1M undeducted contributions this FY.

The Bull said...

Thanks for that.

I wasn't aware that there were such cheap SMSF alternatives available from online providers.

The usual rule of thumb I have seen is that you need $200 - $300k in super to find value in a SMSF (in my opinion closer to $300k! given cost of administration by accounts or SMSF administers). However, with options like this available, you could not argue with investing $100k in a SMSF for an admin fee of 0.75%, if you wanted to be investing in direct equities or Warrants.

For me I will stick with my internally geared share fund in a mastertrust until I have a bigger super balance worth worrying about.

If you do decide to stick with Mastertrusts, have a look around. There's alternatives available that allow you to invest in wholesale funds with no "admin fee" and provide access to index funds with management costs of around 0.40%. They can also offer relatively low cost insurance cover.

From my experience, BT super definitely isn't one of the cheapest going round!